XPK Start: Fri Apr 24 17:26:55 UTC 2026 2026-04-24 17:26:59.271797: 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 00:00:1777051619.284138 9 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777051619.288153 9 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777051619.299591 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777051619.299611 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777051619.299613 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777051619.299615 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 17:27:18.520570: 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) I0424 17:27:19.047456 132604547028800 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 17:27:28,089:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 17:27:28.089562 132604547028800 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 17:27:28,091:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-true-rs-yvo99-slice-job-0-0.mt-15-ocdbt-true-rs-yvo99:8482 I0424 17:27:28.091776 132604547028800 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-yvo99-slice-job-0-0.mt-15-ocdbt-true-rs-yvo99:8482 I0424 17:27:28.974196 132604547028800 max_utils.py:284] Jax distributed system initialized! I0424 17:27:35.891767 132604547028800 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 17:27:35.891875 132604547028800 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 17:27:35.891917 132604547028800 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 I0424 17:27:35.891953 132604547028800 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 17:27:36.516711 132604547028800 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 17:27:36.632273 132604547028800 checkpointing.py:688] Setting up checkpoint logger... I0424 17:27:36.632393 132604547028800 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 17:27:36.632439 132604547028800 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 17:27:36.632652 132604547028800 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 0x7899d3974d10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 17:27:37.895064 132604547028800 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 17:27:37.895398 132604547028800 checkpoint_manager.py:709] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78859824d280>}, handler_registry=None I0424 17:27:37.895685 132604547028800 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78859824d280>` 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`. I0424 17:27:37.895742 132604547028800 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7885985a5d30>` 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`. I0424 17:27:37.895781 132604547028800 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 0x78859824d280>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78859824d280>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7885985a5d30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7885985a5d30>}). I0424 17:27:37.896116 132604547028800 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 17:27:37.896334 132604547028800 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7885983e1260> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 17:27:39.650332 132484977129216 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777051494558859100, 'commit_timestamp_nsecs': 1777051540179378938, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0424 17:27:39.903483 132484977129216 checkpoint_manager.py:1877] Missing metrics for step 9 I0424 17:27:39.942799 132604547028800 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints I0424 17:27:40.294456 132604547028800 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/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78859824c5c0> I0424 17:27:40.294642 132604547028800 checkpointing.py:302] Checkpoint manager created! I0424 17:27:41.248665 132604547028800 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 17:27:41.248785 132604547028800 checkpointing.py:582] restoring from this run's directory step 9 I0424 17:27:41.745584 132604547028800 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9. I0424 17:27:43.146245 1399 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 17:27:50.573247 132604547028800 jax_array_handlers.py:837] [process=6] /jax/orbax/read/worker/io/requested throughput: 212.615 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.4325807094573975 s) (per-host) I0424 17:27:50.581053 132604547028800 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/read/gbytes_per_sec: 202.001 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.834645986557007 s) (per-host) I0424 17:27:50.583258 132604547028800 checkpointer.py:318] Finished restoring checkpoint in 9.09 seconds from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9. I0424 17:27:50.583359 132604547028800 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1777051661.4929152, 'checkpointer_duration_secs': 9.090415954589844, 'checkpoint_manager_start_time': 1777051661.24952, 'checkpoint_manager_duration_secs': 9.333812236785889} fsdp: 32 I0424 17:27:50.663896 132604547028800 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 17:27:50.664014 132604547028800 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 17:27:50.664068 132604547028800 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 17:27:50.664126 132604547028800 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 17:27:50.664168 132604547028800 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 17:27:50.664204 132604547028800 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 17:27:50.664242 132604547028800 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:27:50.664282 132604547028800 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 17:27:50.664328 132604547028800 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:27:50.664365 132604547028800 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:27:50.664402 132604547028800 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 17:27:50.664436 132604547028800 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 17:27:51.063674 132604547028800 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:27:51.063823 132604547028800 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:27:51.069313 132604547028800 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:27:51.069405 132604547028800 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:27:51.224091 132604547028800 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:27:51.224240 132604547028800 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:27:51.378485 132604547028800 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:27:51.378626 132604547028800 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:27:51.543526 132604547028800 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:27:51.543668 132604547028800 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:27:51.694971 132604547028800 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:27:51.695111 132604547028800 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:27:51.844014 132604547028800 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:27:51.844161 132604547028800 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:27:52.004801 132604547028800 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 17:27:52.004936 132604547028800 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:27:52.027728 132604547028800 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 17:27:52.027822 132604547028800 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 17:28:03.261622 132604547028800 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. I0424 17:28:03.264053 132604547028800 metric_logger.py:289] number parameters: 1.104 billion I0424 17:28:14.148781 132604547028800 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 17:28:14.305625 132604547028800 checkpointing.py:798] Waited 0.156829833984375 seconds for step 10 to finish before starting checkpointing. I0424 17:28:14.307939 132604547028800 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:28:14.309765 132604547028800 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0424 17:28:14.312624 132604547028800 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10. I0424 17:28:14.916132 132604547028800 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 17:28:14.917202 132604547028800 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:28:14.917260 132604547028800 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 I0424 17:28:15.170689 132604547028800 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.255710s I0424 17:28:15.170864 132604547028800 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.902 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5315601825714111 s) (per-host) I0424 17:28:15.170917 132604547028800 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.531626s (batch_requests_ready=0.271168s, total_serialization_initiated=0.260384s, others=0.000074s) I0424 17:28:15.171012 132604547028800 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.535863s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=0.535843) I0424 17:28:15.171959 132483903387392 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:28:15.172088 132604547028800 async_checkpointer.py:561] Finished blocking save. Time taken: 0.862252s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10. I0424 17:28:15.194204 132604547028800 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:28:15.194517 132480145291008 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:28:15.194671 132604547028800 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777051694.307921, 'wait_for_prev_duration_secs': 6.365776062011719e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777051694.3098037, 'checkpointer_blocking_duration_secs': 0.8624739646911621, 'get_old_steps_start_time': 1777051695.1722963, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1777051694.3059943, 'checkpoint_manager_blocking_duration_secs': 0.888641357421875} I0424 17:28:15.194780 132604547028800 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 17:28:15.194855 132604547028800 max_utils.py:750] Memstats: After params initialized: I0424 17:28:15.194903 132604547028800 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 17:28:15.194936 132604547028800 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 17:28:15.194963 132604547028800 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 17:28:15.194988 132604547028800 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 17:28:15.467090 132604547028800 metric_logger.py:185] completed step: 10, seconds: 10.884, TFLOP/s/device: 1.248, Tokens/s/device: 188.172, total_weights: 65536, loss: 6.827 I0424 17:28:15.640830 132604547028800 metric_logger.py:185] completed step: 11, seconds: 1.317, TFLOP/s/device: 10.320, Tokens/s/device: 1555.480, total_weights: 65536, loss: 6.522 I0424 17:28:15.788604 132604547028800 metric_logger.py:185] completed step: 12, seconds: 0.019, TFLOP/s/device: 722.681, Tokens/s/device: 108930.376, total_weights: 65536, loss: 6.260 I0424 17:28:15.936113 132604547028800 metric_logger.py:185] completed step: 13, seconds: 0.164, TFLOP/s/device: 82.889, Tokens/s/device: 12493.976, total_weights: 65536, loss: 6.039 I0424 17:28:15.942583 132604547028800 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 17:28:16.231130 132604547028800 checkpointing.py:798] Waited 0.28848958015441895 seconds for step 15 to finish before starting checkpointing. I0424 17:28:16.233579 132604547028800 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:28:20.016885 132480640198400 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_6 I0424 17:28:56.936954 132483903387392 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 37.345 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.29761576652527 s) (per-host) I0424 17:28:56.937072 132483903387392 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.764896s. I0424 17:29:04.700336 132483903387392 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 49.528146s. I0424 17:29:04.700696 132480145291008 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:29:04.700814 132480145291008 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:29:04.700860 132480145291008 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:29:04.702218 132480145291008 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 17:29:04.702386 132604547028800 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0424 17:29:04.702515 132604547028800 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.468938 seconds. If this number is high, consider checkpointing less frequently. I0424 17:29:04.704394 132604547028800 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 15 I0424 17:29:04.708513 132604547028800 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15. I0424 17:29:05.267713 132604547028800 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:29:05.267810 132604547028800 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 I0424 17:29:05.310614 132604547028800 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044926s I0424 17:29:05.310778 132604547028800 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.951 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3115849494934082 s) (per-host) I0424 17:29:05.310827 132604547028800 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.311643s (batch_requests_ready=0.261913s, total_serialization_initiated=0.049665s, others=0.000066s) I0424 17:29:05.310918 132604547028800 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.316408s (all_items=0.000014s, per_item={'items': '0.00001383'}, temp_paths=0.316394) I0424 17:29:05.311775 132480640198400 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:29:05.311931 132604547028800 async_checkpointer.py:561] Finished blocking save. Time taken: 0.607465s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15. I0424 17:29:05.327532 132604547028800 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:29:05.327867 132480145291008 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:29:05.328010 132604547028800 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777051696.2335522, 'wait_for_prev_duration_secs': 48.46893763542175, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777051744.7044334, 'checkpointer_blocking_duration_secs': 0.6076095104217529, 'get_old_steps_start_time': 1777051745.3120608, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1777051696.2314806, 'checkpoint_manager_blocking_duration_secs': 49.096495628356934} I0424 17:29:05.328183 132604547028800 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 17:29:05.328928 132604547028800 metric_logger.py:185] completed step: 14, seconds: 0.146, TFLOP/s/device: 92.806, Tokens/s/device: 13988.689, total_weights: 65536, loss: 5.858 I0424 17:29:05.336256 132604547028800 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.949, Tokens/s/device: 13859.564, total_weights: 65536, loss: 5.713 I0424 17:29:23.445323 132604547028800 metric_logger.py:185] completed step: 16, seconds: 49.392, TFLOP/s/device: 0.275, Tokens/s/device: 41.464, total_weights: 65536, loss: 5.599 I0424 17:29:23.592697 132604547028800 metric_logger.py:185] completed step: 17, seconds: 17.964, TFLOP/s/device: 0.756, Tokens/s/device: 114.007, total_weights: 65536, loss: 5.511 I0424 17:29:23.740129 132604547028800 metric_logger.py:185] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.565, Tokens/s/device: 13349.412, total_weights: 65536, loss: 5.442 I0424 17:29:23.888665 132604547028800 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 17:29:23.889279 132604547028800 checkpointing.py:798] Waited 0.0006248950958251953 seconds for step 19 to finish before starting checkpointing. I0424 17:29:23.891434 132604547028800 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:29:28.210001 132478534678272 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_6 I0424 17:30:04.984162 132480640198400 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 26.333 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 59.98492956161499 s) (per-host) I0424 17:30:04.984290 132480640198400 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 59.672395s. I0424 17:30:11.901228 132480640198400 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 66.589330s. I0424 17:30:11.901566 132480145291008 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:30:11.901680 132480145291008 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:30:11.901727 132480145291008 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:30:11.903328 132480145291008 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 17:30:11.903536 132604547028800 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0424 17:30:11.903604 132604547028800 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.012181 seconds. If this number is high, consider checkpointing less frequently. I0424 17:30:11.905198 132604547028800 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0424 17:30:11.908703 132604547028800 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19. I0424 17:30:12.468172 132604547028800 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:30:12.468284 132604547028800 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 I0424 17:30:12.510567 132604547028800 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044407s I0424 17:30:12.510740 132604547028800 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.060 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3048269748687744 s) (per-host) I0424 17:30:12.510791 132604547028800 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.304888s (batch_requests_ready=0.255866s, total_serialization_initiated=0.048955s, others=0.000067s) I0424 17:30:12.510885 132604547028800 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.308788s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.308776) I0424 17:30:12.511781 132483869816576 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:30:12.511907 132604547028800 async_checkpointer.py:561] Finished blocking save. Time taken: 0.606641s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19. I0424 17:30:12.527387 132604547028800 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:30:12.527666 132480145291008 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:30:12.527819 132604547028800 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777051763.891406, 'wait_for_prev_duration_secs': 48.01218128204346, 'time_between_consecutive_saves_sec': 19.189138650894165, 'checkpointer_blocking_start_time': 1777051811.9052367, 'checkpointer_blocking_duration_secs': 0.6068377494812012, 'get_old_steps_start_time': 1777051812.5120876, 'get_old_steps_duration_secs': 3.552436828613281e-05, 'checkpoint_manager_blocking_start_time': 1777051763.8895023, 'checkpoint_manager_blocking_duration_secs': 48.63828659057617} I0424 17:30:12.527930 132604547028800 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 17:30:12.527972 132604547028800 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:30:17.241575 132483903387392 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_6 I0424 17:30:55.210690 132483869816576 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 36.731 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.004737854003906 s) (per-host) I0424 17:30:55.210806 132483869816576 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.698822s. I0424 17:31:00.444130 132483869816576 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 47.932131s. I0424 17:31:00.444437 132480145291008 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:31:00.444568 132480145291008 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:31:00.444617 132480145291008 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:31:00.445896 132480145291008 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 17:31:00.446072 132604547028800 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0424 17:31:00.446248 132604547028800 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:31:00.446902 132604547028800 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.594, Tokens/s/device: 13956.848, total_weights: 65536, loss: 5.385 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 17:31:10 UTC 2026 EXIT_CODE=0