XPK Start: Tue Apr 21 01:19:16 UTC 2026 2026-04-21 01:19:20.037081: 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 00:00:1776734360.049853 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776734360.053634 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776734360.064663 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776734360.064681 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776734360.064684 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776734360.064685 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 01:19:39.169743: 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 01:19:39.693234 140613597030208 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-21 01:19:48,734:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0421 01:19:48.734833 140613597030208 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-21 01:19:48,737:jax._src.distributed:157: Connecting to JAX distributed service on mt-08-checkpoint-async--pmipe-slice-job-0-0.mt-08-checkpoint-async--pmipe:8482 I0421 01:19:48.737241 140613597030208 distributed.py:157] Connecting to JAX distributed service on mt-08-checkpoint-async--pmipe-slice-job-0-0.mt-08-checkpoint-async--pmipe:8482 I0421 01:19:50.112544 140613597030208 max_utils.py:284] Jax distributed system initialized! I0421 01:19:56.008153 140613597030208 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 01:19:56.008261 140613597030208 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 01:19:56.008303 140613597030208 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 01:19:56.008337 140613597030208 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 01:19:56.619468 140613597030208 maxtext_utils.py:1398] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 01:19:56.619775 140613597030208 checkpointing.py:677] Setting up checkpoint logger... I0421 01:19:56.619831 140613597030208 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 01:19:56.619877 140613597030208 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0421 01:19:56.620240 140613597030208 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 0x7fe29eb6cb90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 01:19:58.299637 140613597030208 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 01:19:58.300003 140613597030208 checkpoint_manager.py:709] [process=4][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe1583e5a00>}, handler_registry=None I0421 01:19:58.300246 140613597030208 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe1583e5a00>` 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 01:19:58.300297 140613597030208 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fd091e46780>` 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 01:19:58.300335 140613597030208 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 0x7fe1583e5a00>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe1583e5a00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fd091e46780>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fd091e46780>}). I0421 01:19:58.303024 140613597030208 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0421 01:19:58.303110 140613597030208 async_checkpointer.py:177] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7fe286ff32e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 01:20:01.019877 140495661545216 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776734232750202128, 'commit_timestamp_nsecs': 1776734279738053398, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_08_checkpoint_async_true/checkpoints/9/_CHECKPOINT_METADATA I0421 01:20:01.271788 140495661545216 checkpoint_manager.py:1877] Missing metrics for step 9 I0421 01:20:02.114387 140613597030208 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_08_checkpoint_async_true/checkpoints I0421 01:20:03.409466 140613597030208 checkpoint_manager.py:929] [process=4][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_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fd091e403b0> I0421 01:20:03.409644 140613597030208 checkpointing.py:301] Checkpoint manager created! I0421 01:20:04.293499 140613597030208 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 01:20:04.293601 140613597030208 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 01:20:04.665546 140613597030208 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 01:20:04.665641 140613597030208 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 01:20:04.681285 140613597030208 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 01:20:04.681346 140613597030208 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 01:20:04.704580 140613597030208 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 01:20:04.704648 140613597030208 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 01:20:04.720392 140613597030208 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 01:20:04.720456 140613597030208 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 01:20:04.736074 140613597030208 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 01:20:04.736135 140613597030208 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 01:20:04.759840 140613597030208 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 01:20:04.759912 140613597030208 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 01:20:04.779256 140613597030208 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 01:20:04.779319 140613597030208 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 01:20:04.975110 140613597030208 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 01:20:04.975220 140613597030208 checkpointing.py:581] restoring from this run's directory step 9 I0421 01:20:05.480311 140613597030208 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_08_checkpoint_async_true/checkpoints/9. I0421 01:20:07.283111 1370 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 01:20:13.750203 140613597030208 jax_array_handlers.py:837] [process=4] /jax/orbax/read/worker/io/requested throughput: 244.158 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.472336292266846 s) (per-host) I0421 01:20:13.763128 140613597030208 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/read/gbytes_per_sec: 230.162 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.876013517379761 s) (per-host) I0421 01:20:16.194087 140613597030208 checkpointer.py:318] Finished restoring checkpoint in 10.96 seconds from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_08_checkpoint_async_true/checkpoints/9. I0421 01:20:16.194247 140613597030208 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_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776734405.2303524, 'checkpointer_duration_secs': 10.963863611221313, 'checkpoint_manager_start_time': 1776734404.9754598, 'checkpoint_manager_duration_secs': 11.218758344650269} fsdp: 32 I0421 01:20:16.597434 140613597030208 maxtext_utils.py:1501] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 01:20:16.597565 140613597030208 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 01:20:16.597621 140613597030208 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 01:20:16.597680 140613597030208 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 01:20:16.597752 140613597030208 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 01:20:16.597795 140613597030208 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 01:20:16.597851 140613597030208 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 01:20:16.597923 140613597030208 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 01:20:16.597971 140613597030208 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 01:20:16.598011 140613597030208 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 01:20:16.598060 140613597030208 maxtext_utils.py:1501] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0421 01:20:16.598111 140613597030208 maxtext_utils.py:1501] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0421 01:20:17.098005 140613597030208 train.py:156] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 01:20:17.098098 140613597030208 train.py:156] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0421 01:20:17.112900 140613597030208 train.py:163] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 01:20:17.112959 140613597030208 train.py:163] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0421 01:20:27.959632 140613597030208 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 01:20:27.960545 140613597030208 metric_logger.py:289] number parameters: 1.104 billion I0421 01:20:39.042149 140613597030208 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 01:20:39.195054 140613597030208 checkpointing.py:776] Waited 0.15288662910461426 seconds for step 10 to finish before starting checkpointing. I0421 01:20:39.197646 140613597030208 checkpoint_manager.py:2013] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 01:20:39.199543 140613597030208 checkpoint_manager.py:1518] [process=4] Saving checkpoint at step 10 I0421 01:20:39.202311 140613597030208 async_checkpointer.py:452] [process=4] 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_08_checkpoint_async_true/checkpoints/10. I0421 01:20:40.616008 140613597030208 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 01:20:40.617271 140613597030208 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 01:20:40.617327 140613597030208 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 01:20:40.866372 140613597030208 base_pytree_checkpoint_handler.py:153] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.250457s I0421 01:20:40.866550 140613597030208 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.658 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.9304237365722656 s) (per-host) I0421 01:20:40.866602 140613597030208 base_pytree_checkpoint_handler.py:737] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.930487s (batch_requests_ready=0.677937s, total_serialization_initiated=0.252478s, others=0.000072s) I0421 01:20:40.866709 140613597030208 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.934656s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.934638) I0421 01:20:40.867640 140490296624896 async_checkpointer.py:79] [process=4][thread=async_save] Background save thread started. I0421 01:20:40.867818 140613597030208 async_checkpointer.py:561] Finished blocking save. Time taken: 1.668201s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_08_checkpoint_async_true/checkpoints/10. I0421 01:20:40.915547 140613597030208 checkpoint_manager.py:1566] [process=4][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 01:20:40.915973 140492432803584 async_checkpointer.py:265] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 01:20:40.916137 140613597030208 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776734439.197626, 'wait_for_prev_duration_secs': 6.556510925292969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776734439.1995842, 'checkpointer_blocking_duration_secs': 1.6683368682861328, 'get_old_steps_start_time': 1776734440.8679383, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776734439.1956327, 'checkpoint_manager_blocking_duration_secs': 1.7204608917236328} I0421 01:20:40.916243 140613597030208 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 01:20:40.916294 140613597030208 max_utils.py:750] Memstats: After params initialized: I0421 01:20:40.916346 140613597030208 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0)) I0421 01:20:40.916380 140613597030208 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0)) I0421 01:20:40.916408 140613597030208 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0)) I0421 01:20:40.916438 140613597030208 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0)) I0421 01:20:41.176576 140613597030208 metric_logger.py:185] completed step: 10, seconds: 11.081, TFLOP/s/device: 1.226, Tokens/s/device: 184.813, total_weights: 65536, loss: 6.833 I0421 01:20:41.357791 140613597030208 metric_logger.py:185] completed step: 11, seconds: 2.133, TFLOP/s/device: 6.369, Tokens/s/device: 960.073, total_weights: 65536, loss: 6.528 I0421 01:20:41.505599 140613597030208 metric_logger.py:185] completed step: 12, seconds: 0.034, TFLOP/s/device: 396.230, Tokens/s/device: 59724.126, total_weights: 65536, loss: 6.265 I0421 01:20:41.653141 140613597030208 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 89.028, Tokens/s/device: 13419.213, total_weights: 65536, loss: 6.045 I0421 01:20:41.657515 140613597030208 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 01:20:41.947728 140613597030208 checkpointing.py:776] Waited 0.29015636444091797 seconds for step 15 to finish before starting checkpointing. I0421 01:20:41.950483 140613597030208 checkpoint_manager.py:2024] [process=4][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 01:20:47.041632 140490863277824 array_metadata_store.py:203] [process=4][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_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_4 I0421 01:21:24.081377 140490296624896 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/gbytes_per_sec: 35.782 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.145214796066284 s) (per-host) I0421 01:21:24.081504 140490296624896 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.213743s. I0421 01:21:31.340439 140490296624896 async_checkpointer.py:144] [process=4][thread=async_save] Background save thread done. Time taken: 50.472662s. I0421 01:21:31.340747 140492432803584 async_checkpointer.py:273] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 01:21:31.340866 140492432803584 async_checkpointer.py:283] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 01:21:31.340915 140492432803584 checkpoint_manager.py:2133] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 01:21:31.342556 140492432803584 checkpoint_manager.py:2142] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 01:21:31.342742 140613597030208 checkpoint_manager.py:2036] [process=4][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0421 01:21:31.342873 140613597030208 checkpoint_manager.py:1458] Waiting for previous save to complete took 49.392393 seconds. If this number is high, consider checkpointing less frequently. I0421 01:21:31.344715 140613597030208 checkpoint_manager.py:1518] [process=4] Saving checkpoint at step 15 I0421 01:21:31.348834 140613597030208 async_checkpointer.py:452] [process=4] 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_08_checkpoint_async_true/checkpoints/15. I0421 01:21:32.304114 140613597030208 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 01:21:32.304218 140613597030208 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 01:21:32.344573 140613597030208 base_pytree_checkpoint_handler.py:153] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041600s I0421 01:21:32.344748 140613597030208 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.255 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2935338020324707 s) (per-host) I0421 01:21:32.344804 140613597030208 base_pytree_checkpoint_handler.py:737] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.293600s (batch_requests_ready=0.249928s, total_serialization_initiated=0.043599s, others=0.000073s) I0421 01:21:32.344898 140613597030208 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.298541s (all_items=0.000012s, per_item={'items': '0.00001192'}, temp_paths=0.298529) I0421 01:21:32.345751 140491374970624 async_checkpointer.py:79] [process=4][thread=async_save] Background save thread started. I0421 01:21:32.345911 140613597030208 async_checkpointer.py:561] Finished blocking save. Time taken: 1.001124s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_08_checkpoint_async_true/checkpoints/15. I0421 01:21:32.364326 140613597030208 checkpoint_manager.py:1566] [process=4][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 01:21:32.364665 140492432803584 async_checkpointer.py:265] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 01:21:32.364846 140613597030208 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776734441.9504557, 'wait_for_prev_duration_secs': 49.392393350601196, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776734491.3447545, 'checkpointer_blocking_duration_secs': 1.0012643337249756, 'get_old_steps_start_time': 1776734492.346038, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776734441.9483333, 'checkpoint_manager_blocking_duration_secs': 50.41648006439209} I0421 01:21:32.364953 140613597030208 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0421 01:21:32.365474 140613597030208 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.058, Tokens/s/device: 13875.997, total_weights: 65536, loss: 5.864 I0421 01:21:32.371115 140613597030208 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 92.051, Tokens/s/device: 13874.869, total_weights: 65536, loss: 5.719 I0421 01:21:50.470237 140613597030208 metric_logger.py:185] completed step: 16, seconds: 50.712, TFLOP/s/device: 0.268, Tokens/s/device: 40.385, total_weights: 65536, loss: 5.605 I0421 01:21:50.617828 140613597030208 metric_logger.py:185] completed step: 17, seconds: 17.953, TFLOP/s/device: 0.757, Tokens/s/device: 114.075, total_weights: 65536, loss: 5.517 I0421 01:21:50.765339 140613597030208 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.563, Tokens/s/device: 13499.974, total_weights: 65536, loss: 5.448 I0421 01:21:50.912327 140613597030208 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 01:21:50.912935 140613597030208 checkpointing.py:776] Waited 0.0006289482116699219 seconds for step 19 to finish before starting checkpointing. I0421 01:21:50.915016 140613597030208 checkpoint_manager.py:2024] [process=4][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 01:21:57.025125 140490863277824 array_metadata_store.py:203] [process=4][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_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_4 I0421 01:22:33.307782 140491374970624 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/gbytes_per_sec: 25.787 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 61.25653624534607 s) (per-host) I0421 01:22:33.307887 140491374970624 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 60.962027s. I0421 01:22:40.468711 140491374970624 async_checkpointer.py:144] [process=4][thread=async_save] Background save thread done. Time taken: 68.122816s. I0421 01:22:40.468985 140492432803584 async_checkpointer.py:273] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 01:22:40.469111 140492432803584 async_checkpointer.py:283] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 01:22:40.469168 140492432803584 checkpoint_manager.py:2133] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 01:22:40.470487 140492432803584 checkpoint_manager.py:2142] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 01:22:40.470667 140613597030208 checkpoint_manager.py:2036] [process=4][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0421 01:22:40.470801 140613597030208 checkpoint_manager.py:1458] Waiting for previous save to complete took 49.555798 seconds. If this number is high, consider checkpointing less frequently. I0421 01:22:40.472188 140613597030208 checkpoint_manager.py:1518] [process=4] Saving checkpoint at step 19 I0421 01:22:40.475555 140613597030208 async_checkpointer.py:452] [process=4] 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_08_checkpoint_async_true/checkpoints/19. I0421 01:22:42.281662 140613597030208 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0421 01:22:42.281770 140613597030208 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 01:22:42.321304 140613597030208 base_pytree_checkpoint_handler.py:153] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040821s I0421 01:22:42.321482 140613597030208 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.136 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7223453521728516 s) (per-host) I0421 01:22:42.321534 140613597030208 base_pytree_checkpoint_handler.py:737] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.722408s (batch_requests_ready=0.679482s, total_serialization_initiated=0.042857s, others=0.000069s) I0421 01:22:42.321624 140613597030208 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.726344s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.726332) I0421 01:22:42.322510 140492424410880 async_checkpointer.py:79] [process=4][thread=async_save] Background save thread started. I0421 01:22:42.322666 140613597030208 async_checkpointer.py:561] Finished blocking save. Time taken: 1.850410s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260421_005327/linen_xpk_test_pipeline_scan_nnx_20260421_005327_08_checkpoint_async_true/checkpoints/19. I0421 01:22:42.335021 140613597030208 checkpoint_manager.py:1566] [process=4][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 01:22:42.335252 140492432803584 async_checkpointer.py:265] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 01:22:42.335376 140613597030208 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776734510.9149876, 'wait_for_prev_duration_secs': 49.5557975769043, 'time_between_consecutive_saves_sec': 19.572388887405396, 'checkpointer_blocking_start_time': 1776734560.472226, 'checkpointer_blocking_duration_secs': 1.8505582809448242, 'get_old_steps_start_time': 1776734562.322796, 'get_old_steps_duration_secs': 2.1696090698242188e-05, 'checkpoint_manager_blocking_start_time': 1776734510.9131649, 'checkpoint_manager_blocking_duration_secs': 51.422178745269775} I0421 01:22:42.335484 140613597030208 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 01:22:42.335526 140613597030208 checkpoint_manager.py:2024] [process=4][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 01:22:48.102588 140495149852416 array_metadata_store.py:203] [process=4][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_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_4 I0421 01:23:25.538109 140492424410880 base_pytree_checkpoint_handler.py:129] [process=4] /jax/checkpoint/write/gbytes_per_sec: 35.950 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.938947439193726 s) (per-host) I0421 01:23:25.538206 140492424410880 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.215587s. I0421 01:23:32.512500 140492424410880 async_checkpointer.py:144] [process=4][thread=async_save] Background save thread done. Time taken: 50.189865s. I0421 01:23:32.512758 140492432803584 async_checkpointer.py:273] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 01:23:32.512817 140492432803584 async_checkpointer.py:283] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 01:23:32.512855 140492432803584 checkpoint_manager.py:2133] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 01:23:32.514656 140492432803584 checkpoint_manager.py:2142] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 01:23:32.514842 140613597030208 checkpoint_manager.py:2036] [process=4][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0421 01:23:32.514990 140613597030208 checkpoint_manager.py:2013] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 01:23:32.515648 140613597030208 metric_logger.py:185] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.097, Tokens/s/device: 13881.828, 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 01:23:41 UTC 2026 EXIT_CODE=0