XPK Start: Fri Apr 24 17:50:15 UTC 2026 2026-04-24 17:50:19.580909: 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 00:00:1777053019.593805 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777053019.597769 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777053019.609481 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777053019.609501 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777053019.609503 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777053019.609505 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 17:50:38.834528: 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 17:50:39.367840 138452135364416 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 17:50:48,408:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 17:50:48.408847 138452135364416 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 17:50:48,411:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-false-sv-4sgxz-slice-job-0-0.mt-16-zarr3-false-sv-4sgxz:8482 I0424 17:50:48.411277 138452135364416 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-false-sv-4sgxz-slice-job-0-0.mt-16-zarr3-false-sv-4sgxz:8482 I0424 17:50:49.487536 138452135364416 max_utils.py:284] Jax distributed system initialized! I0424 17:50:56.579366 138452135364416 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 17:50:56.579475 138452135364416 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 17:50:56.579519 138452135364416 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 17:50:56.579554 138452135364416 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 17:50:57.200237 138452135364416 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 17:50:57.314529 138452135364416 checkpointing.py:688] Setting up checkpoint logger... I0424 17:50:57.314644 138452135364416 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0424 17:50:57.314689 138452135364416 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 17:50:57.314905 138452135364416 base_pytree_checkpoint_handler.py:415] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7deb64775760>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 17:51:00.199275 138452135364416 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 17:51:00.199519 138452135364416 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 0x7dd73413ff50>}, handler_registry=None I0424 17:51:00.199762 138452135364416 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dd73413ff50>` 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 17:51:00.199812 138452135364416 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dd7344773b0>` 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 17:51:00.199848 138452135364416 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 0x7dd73413ff50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dd73413ff50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dd7344773b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dd7344773b0>}). I0424 17:51:00.200205 138452135364416 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 17:51:00.200289 138452135364416 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7dd764118f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 17:51:01.232821 138452135364416 checkpoint_manager.py:1818] Found 0 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_16_zarr3_false/checkpoints I0424 17:51:01.269026 138452135364416 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_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dd73413c5f0> I0424 17:51:01.269188 138452135364416 checkpointing.py:302] Checkpoint manager created! I0424 17:51:02.207150 138452135364416 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 17:51:02.207276 138452135364416 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 17:51:03.621001 138452135364416 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 17:51:03.621138 138452135364416 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 17:51:03.621196 138452135364416 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 17:51:03.621241 138452135364416 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 17:51:03.621281 138452135364416 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 17:51:03.621318 138452135364416 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 17:51:03.621356 138452135364416 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:51:03.621394 138452135364416 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 17:51:03.621430 138452135364416 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:51:03.621465 138452135364416 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 17:51:03.621504 138452135364416 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 17:51:03.621540 138452135364416 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 17:51:04.017652 138452135364416 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:51:04.017803 138452135364416 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:51:04.023244 138452135364416 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:51:04.023338 138452135364416 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:51:04.172089 138452135364416 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:51:04.172240 138452135364416 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:51:04.316936 138452135364416 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:51:04.317070 138452135364416 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 17:51:04.468260 138452135364416 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 17:51:04.468399 138452135364416 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:51:04.608586 138452135364416 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 17:51:04.608721 138452135364416 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:51:04.747956 138452135364416 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 17:51:04.748090 138452135364416 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:51:04.897643 138452135364416 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 17:51:04.897778 138452135364416 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 17:51:04.921583 138452135364416 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 17:51:04.921684 138452135364416 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 17:51:16.012932 138452135364416 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 17:51:16.015469 138452135364416 metric_logger.py:289] number parameters: 1.104 billion I0424 17:51:27.397883 138452135364416 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 17:51:27.549137 138452135364416 checkpointing.py:798] Waited 0.15120649337768555 seconds for step 0 to finish before starting checkpointing. I0424 17:51:27.551510 138452135364416 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:51:27.553162 138452135364416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0424 17:51:27.555915 138452135364416 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_16_zarr3_false/checkpoints/0. I0424 17:51:28.298220 138452135364416 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 17:51:28.299279 138452135364416 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:51:28.299345 138452135364416 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 17:51:28.547990 138452135364416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.250902s I0424 17:51:28.548185 138452135364416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.952 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5225205421447754 s) (per-host) I0424 17:51:28.548244 138452135364416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.522592s (batch_requests_ready=0.265119s, total_serialization_initiated=0.257393s, others=0.000080s) I0424 17:51:28.548341 138452135364416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.526588s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.526569) I0424 17:51:28.549363 138326929565440 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:51:28.549543 138452135364416 async_checkpointer.py:561] Finished blocking save. Time taken: 0.996303s. 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_16_zarr3_false/checkpoints/0. I0424 17:51:28.583809 138452135364416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:51:28.584136 138325864216320 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:51:28.584291 138452135364416 standard_logger.py:34] {'step': 0, '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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777053087.5514915, 'wait_for_prev_duration_secs': 6.580352783203125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777053087.5532017, 'checkpointer_blocking_duration_secs': 0.996450662612915, 'get_old_steps_start_time': 1777053088.5496733, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1777053087.5495422, 'checkpoint_manager_blocking_duration_secs': 1.0347027778625488} I0424 17:51:28.584467 138452135364416 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 17:51:28.584547 138452135364416 max_utils.py:750] Memstats: After params initialized: I0424 17:51:28.584601 138452135364416 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 17:51:28.584636 138452135364416 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 17:51:28.584664 138452135364416 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 17:51:28.584690 138452135364416 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 17:51:28.856115 138452135364416 metric_logger.py:185] completed step: 0, seconds: 11.381, TFLOP/s/device: 1.194, Tokens/s/device: 179.944, total_weights: 65536, loss: 10.874 I0424 17:51:29.020814 138452135364416 metric_logger.py:185] completed step: 1, seconds: 1.456, TFLOP/s/device: 9.329, Tokens/s/device: 1406.185, total_weights: 65536, loss: 10.874 I0424 17:51:29.446817 138452135364416 metric_logger.py:185] completed step: 2, seconds: 0.018, TFLOP/s/device: 735.194, Tokens/s/device: 110816.514, total_weights: 65536, loss: 10.560 I0424 17:51:29.594218 138452135364416 metric_logger.py:185] completed step: 3, seconds: 0.417, TFLOP/s/device: 32.569, Tokens/s/device: 4909.128, total_weights: 65536, loss: 9.979 I0424 17:51:29.600052 138452135364416 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0424 17:51:29.889240 138452135364416 checkpointing.py:798] Waited 0.28915858268737793 seconds for step 5 to finish before starting checkpointing. I0424 17:51:29.891765 138452135364416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:51:31.322768 2578 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 17:51:33.223051 138326937958144 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_16_zarr3_false/checkpoints/0/items/array_metadatas/process_6 I0424 17:52:09.180862 138326929565440 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 38.382 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.1551399230957 s) (per-host) I0424 17:52:09.180982 138326929565440 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.631507s. I0424 17:52:18.551692 138326929565440 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 50.002201s. I0424 17:52:18.551990 138325864216320 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:52:18.552138 138325864216320 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:52:18.552210 138325864216320 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:52:18.554049 138325864216320 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 17:52:18.554244 138452135364416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0424 17:52:18.554384 138452135364416 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.662623 seconds. If this number is high, consider checkpointing less frequently. I0424 17:52:18.556334 138452135364416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 5 I0424 17:52:18.560463 138452135364416 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_16_zarr3_false/checkpoints/5. I0424 17:52:19.149631 138452135364416 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:52:19.149786 138452135364416 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 17:52:19.194420 138452135364416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046830s I0424 17:52:19.194582 138452135364416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.102 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30234289169311523 s) (per-host) I0424 17:52:19.194632 138452135364416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.302402s (batch_requests_ready=0.250392s, total_serialization_initiated=0.051943s, others=0.000067s) I0424 17:52:19.194723 138452135364416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.307336s (all_items=0.000015s, per_item={'items': '0.00001454'}, temp_paths=0.307322) I0424 17:52:19.195637 138326937958144 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:52:19.195770 138452135364416 async_checkpointer.py:561] Finished blocking save. Time taken: 0.639364s. 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_16_zarr3_false/checkpoints/5. I0424 17:52:19.225878 138452135364416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:52:19.226180 138325864216320 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:52:19.226337 138452135364416 standard_logger.py:34] {'step': 5, '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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777053089.8917375, 'wait_for_prev_duration_secs': 48.66262340545654, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777053138.5563726, 'checkpointer_blocking_duration_secs': 0.6395721435546875, 'get_old_steps_start_time': 1777053139.1959646, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1777053089.8895967, 'checkpoint_manager_blocking_duration_secs': 49.33670425415039} I0424 17:52:19.226448 138452135364416 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0424 17:52:19.227227 138452135364416 metric_logger.py:185] completed step: 4, seconds: 0.163, TFLOP/s/device: 83.259, Tokens/s/device: 12549.712, total_weights: 65536, loss: 9.457 I0424 17:52:19.234378 138452135364416 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.518, Tokens/s/device: 13945.349, total_weights: 65536, loss: 8.955 I0424 17:52:39.053380 138452135364416 metric_logger.py:185] completed step: 6, seconds: 49.633, TFLOP/s/device: 0.274, Tokens/s/device: 41.263, total_weights: 65536, loss: 8.465 I0424 17:52:39.200778 138452135364416 metric_logger.py:185] completed step: 7, seconds: 19.674, TFLOP/s/device: 0.691, Tokens/s/device: 104.099, total_weights: 65536, loss: 7.998 I0424 17:52:39.348188 138452135364416 metric_logger.py:185] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.623, Tokens/s/device: 13358.293, total_weights: 65536, loss: 7.567 I0424 17:52:39.496643 138452135364416 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 17:52:39.497257 138452135364416 checkpointing.py:798] Waited 0.0006265640258789062 seconds for step 9 to finish before starting checkpointing. I0424 17:52:39.499238 138452135364416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:52:43.682285 138326384301824 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_16_zarr3_false/checkpoints/5/items/array_metadatas/process_6 I0424 17:53:20.343707 138326937958144 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 25.705 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 61.45142388343811 s) (per-host) I0424 17:53:20.343824 138326937958144 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 61.147972s. I0424 17:53:28.716650 138326937958144 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 69.520783s. I0424 17:53:28.716879 138325864216320 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:53:28.716937 138325864216320 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:53:28.716990 138325864216320 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:53:28.718535 138325864216320 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0424 17:53:28.718736 138452135364416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0424 17:53:28.718862 138452135364416 checkpoint_manager.py:1458] Waiting for previous save to complete took 49.219633 seconds. If this number is high, consider checkpointing less frequently. I0424 17:53:28.720322 138452135364416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0424 17:53:28.723847 138452135364416 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_16_zarr3_false/checkpoints/9. I0424 17:53:29.283684 138452135364416 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:53:29.283782 138452135364416 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 17:53:29.328836 138452135364416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.047127s I0424 17:53:29.329002 138452135364416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.927 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.31311750411987305 s) (per-host) I0424 17:53:29.329050 138452135364416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.313175s (batch_requests_ready=0.261459s, total_serialization_initiated=0.051654s, others=0.000062s) I0424 17:53:29.329165 138452135364416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.317145s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.317134) I0424 17:53:29.330029 138326895994624 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:53:29.330166 138452135364416 async_checkpointer.py:561] Finished blocking save. Time taken: 0.609773s. 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_16_zarr3_false/checkpoints/9. I0424 17:53:29.333702 138452135364416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:53:29.333934 138325864216320 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:53:29.334087 138452135364416 standard_logger.py:34] {'step': 9, '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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777053159.499212, 'wait_for_prev_duration_secs': 49.21963334083557, 'time_between_consecutive_saves_sec': 20.945122241973877, 'checkpointer_blocking_start_time': 1777053208.720361, 'checkpointer_blocking_duration_secs': 0.6099684238433838, 'get_old_steps_start_time': 1777053209.3303428, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1777053159.4974797, 'checkpoint_manager_blocking_duration_secs': 49.836575984954834} I0424 17:53:29.334214 138452135364416 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 17:53:29.334263 138452135364416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:53:34.156160 138326929565440 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_16_zarr3_false/checkpoints/9/items/array_metadatas/process_6 I0424 17:54:10.297497 138326895994624 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 38.264 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.28157353401184 s) (per-host) I0424 17:54:10.297616 138326895994624 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.967370s. I0424 17:54:28.926968 138326895994624 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 59.596708s. I0424 17:54:28.927289 138325864216320 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:54:28.927417 138325864216320 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:54:28.927459 138325864216320 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:54:28.929003 138325864216320 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 17:54:28.929193 138452135364416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0424 17:54:28.929355 138452135364416 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:54:28.929974 138452135364416 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.512, Tokens/s/device: 13944.399, total_weights: 65536, loss: 7.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 17:54:41 UTC 2026 EXIT_CODE=0