XPK Start: Thu Apr 23 18:34:43 UTC 2026 2026-04-23 18:34:47.859553: 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 00:00:1776969287.872504 9 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776969287.876259 9 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776969287.887916 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969287.887935 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969287.887938 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969287.887940 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 18:35:07.235678: 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) I0423 18:35:07.762969 135932142696256 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 18:35:16,804:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 18:35:16.804184 135932142696256 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 18:35:16,806:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-true-rs-16wtd-slice-job-0-0.mt-15-ocdbt-true-rs-16wtd:8482 I0423 18:35:16.806546 135932142696256 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-16wtd-slice-job-0-0.mt-15-ocdbt-true-rs-16wtd:8482 I0423 18:35:17.747312 135932142696256 max_utils.py:284] Jax distributed system initialized! I0423 18:35:24.571649 135932142696256 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 18:35:24.571753 135932142696256 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 18:35:24.571794 135932142696256 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 I0423 18:35:24.571829 135932142696256 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 18:35:25.191252 135932142696256 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 18:35:25.305800 135932142696256 checkpointing.py:688] Setting up checkpoint logger... I0423 18:35:25.305918 135932142696256 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 18:35:25.305963 135932142696256 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 18:35:25.306184 135932142696256 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 0x7ba0b3b79250>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 18:35:26.746835 135932142696256 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 18:35:26.747192 135932142696256 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 0x7b8c80477e00>}, handler_registry=None I0423 18:35:26.747436 135932142696256 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b8c80477e00>` 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`. I0423 18:35:26.747485 135932142696256 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b8c80474bf0>` 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`. I0423 18:35:26.747522 135932142696256 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 0x7b8c80477e00>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b8c80477e00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b8c80474bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b8c80474bf0>}). I0423 18:35:26.747851 135932142696256 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 18:35:26.747930 135932142696256 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b8d5a4c4f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 18:35:28.947194 135811882276608 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776969162891549598, 'commit_timestamp_nsecs': 1776969208061134700, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0423 18:35:29.196516 135811882276608 checkpoint_manager.py:1877] Missing metrics for step 9 I0423 18:35:29.422076 135932142696256 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints I0423 18:35:29.683739 135932142696256 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b8c800985c0> I0423 18:35:29.683918 135932142696256 checkpointing.py:302] Checkpoint manager created! I0423 18:35:30.620635 135932142696256 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 18:35:30.620753 135932142696256 checkpointing.py:582] restoring from this run's directory step 9 I0423 18:35:31.107822 135932142696256 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/9. I0423 18:35:32.570238 1414 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 18:35:38.807547 135932142696256 jax_array_handlers.py:837] [process=6] /jax/orbax/read/worker/io/requested throughput: 253.127 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.243005990982056 s) (per-host) I0423 18:35:38.816260 135932142696256 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/read/gbytes_per_sec: 236.376 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.695286273956299 s) (per-host) I0423 18:35:41.939986 135932142696256 checkpointer.py:318] Finished restoring checkpoint in 11.07 seconds from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/9. I0423 18:35:41.940184 135932142696256 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776969330.86527, 'checkpointer_duration_secs': 11.074882745742798, 'checkpoint_manager_start_time': 1776969330.621359, 'checkpoint_manager_duration_secs': 11.318795680999756} fsdp: 32 I0423 18:35:42.021509 135932142696256 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 18:35:42.021632 135932142696256 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 18:35:42.021687 135932142696256 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 18:35:42.021731 135932142696256 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 18:35:42.021770 135932142696256 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 18:35:42.021806 135932142696256 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 18:35:42.021843 135932142696256 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:35:42.021879 135932142696256 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 18:35:42.021914 135932142696256 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:35:42.021948 135932142696256 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:35:42.021981 135932142696256 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 18:35:42.022013 135932142696256 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 18:35:42.418771 135932142696256 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 18:35:42.418922 135932142696256 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:35:42.424334 135932142696256 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 18:35:42.424433 135932142696256 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:35:42.582246 135932142696256 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 18:35:42.582393 135932142696256 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:35:42.736233 135932142696256 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 18:35:42.736371 135932142696256 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:35:42.902400 135932142696256 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 18:35:42.902547 135932142696256 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:35:43.059043 135932142696256 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 18:35:43.059195 135932142696256 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:35:43.215493 135932142696256 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 18:35:43.215627 135932142696256 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:35:43.380384 135932142696256 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 18:35:43.380523 135932142696256 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:35:43.403238 135932142696256 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 18:35:43.403335 135932142696256 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 18:35:54.504841 135932142696256 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. I0423 18:35:54.507400 135932142696256 metric_logger.py:289] number parameters: 1.104 billion I0423 18:36:05.753664 135932142696256 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0423 18:36:05.904393 135932142696256 checkpointing.py:798] Waited 0.15071725845336914 seconds for step 10 to finish before starting checkpointing. I0423 18:36:05.906651 135932142696256 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 18:36:05.908597 135932142696256 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0423 18:36:05.911252 135932142696256 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/10. I0423 18:36:06.937434 135932142696256 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 18:36:06.938612 135932142696256 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:36:06.938672 135932142696256 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 I0423 18:36:07.182973 135932142696256 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.246663s I0423 18:36:07.183153 135932142696256 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.650 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.9348146915435791 s) (per-host) I0423 18:36:07.183205 135932142696256 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.934877s (batch_requests_ready=0.683430s, total_serialization_initiated=0.251377s, others=0.000070s) I0423 18:36:07.183303 135932142696256 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.938907s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.938889) I0423 18:36:07.184221 135811286689536 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:36:07.184394 135932142696256 async_checkpointer.py:561] Finished blocking save. Time taken: 1.275725s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/10. I0423 18:36:07.188909 135932142696256 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:36:07.189170 135810716247808 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:36:07.189325 135932142696256 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969365.906634, 'wait_for_prev_duration_secs': 6.318092346191406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776969365.9086359, 'checkpointer_blocking_duration_secs': 1.2758667469024658, 'get_old_steps_start_time': 1776969367.184523, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776969365.9047632, 'checkpoint_manager_blocking_duration_secs': 1.2845280170440674} I0423 18:36:07.189435 135932142696256 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0423 18:36:07.189506 135932142696256 max_utils.py:750] Memstats: After params initialized: I0423 18:36:07.189554 135932142696256 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 18:36:07.189587 135932142696256 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 18:36:07.189615 135932142696256 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 18:36:07.189640 135932142696256 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 18:36:07.462847 135932142696256 metric_logger.py:185] completed step: 10, seconds: 11.245, TFLOP/s/device: 1.208, Tokens/s/device: 182.122, total_weights: 65536, loss: 6.827 I0423 18:36:07.637431 135932142696256 metric_logger.py:185] completed step: 11, seconds: 1.708, TFLOP/s/device: 7.957, Tokens/s/device: 1199.409, total_weights: 65536, loss: 6.522 I0423 18:36:07.784804 135932142696256 metric_logger.py:185] completed step: 12, seconds: 0.019, TFLOP/s/device: 710.327, Tokens/s/device: 107068.172, total_weights: 65536, loss: 6.260 I0423 18:36:07.932138 135932142696256 metric_logger.py:185] completed step: 13, seconds: 0.163, TFLOP/s/device: 83.147, Tokens/s/device: 12532.893, total_weights: 65536, loss: 6.039 I0423 18:36:07.939549 135932142696256 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0423 18:36:08.227064 135932142696256 checkpointing.py:798] Waited 0.28748512268066406 seconds for step 15 to finish before starting checkpointing. I0423 18:36:08.230456 135932142696256 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:36:12.435976 135805926373120 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_6 I0423 18:36:49.243963 135811286689536 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 36.739 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.995588302612305 s) (per-host) I0423 18:36:49.244086 135811286689536 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.059749s. I0423 18:36:55.655078 135811286689536 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 48.470727s. I0423 18:36:55.655424 135810716247808 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:36:55.655553 135810716247808 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:36:55.655601 135810716247808 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:36:55.656887 135810716247808 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 18:36:55.657058 135932142696256 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. W0423 18:36:55.657202 135932142696256 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.426754 seconds. If this number is high, consider checkpointing less frequently. I0423 18:36:55.659356 135932142696256 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 15 I0423 18:36:55.663736 135932142696256 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/15. I0423 18:36:56.294000 135932142696256 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:36:56.294110 135932142696256 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 I0423 18:36:56.335832 135932142696256 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044163s I0423 18:36:56.335996 135932142696256 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.184 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29758262634277344 s) (per-host) I0423 18:36:56.336050 135932142696256 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.297647s (batch_requests_ready=0.248743s, total_serialization_initiated=0.048832s, others=0.000072s) I0423 18:36:56.336181 135932142696256 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.302504s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.302493) I0423 18:36:56.337007 135806983329536 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:36:56.337146 135932142696256 async_checkpointer.py:561] Finished blocking save. Time taken: 0.677720s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/15. I0423 18:36:56.355550 135932142696256 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:36:56.355863 135810716247808 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:36:56.356021 135932142696256 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969368.230425, 'wait_for_prev_duration_secs': 47.426753520965576, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776969415.6593957, 'checkpointer_blocking_duration_secs': 0.6779406070709229, 'get_old_steps_start_time': 1776969416.337355, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776969368.2274704, 'checkpoint_manager_blocking_duration_secs': 48.1285183429718} I0423 18:36:56.356151 135932142696256 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0423 18:36:56.356887 135932142696256 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.477, Tokens/s/device: 13939.179, total_weights: 65536, loss: 5.858 I0423 18:36:56.364278 135932142696256 metric_logger.py:185] completed step: 15, seconds: 0.149, TFLOP/s/device: 91.394, Tokens/s/device: 13775.939, total_weights: 65536, loss: 5.713 I0423 18:37:15.010591 135932142696256 metric_logger.py:185] completed step: 16, seconds: 48.423, TFLOP/s/device: 0.281, Tokens/s/device: 42.294, total_weights: 65536, loss: 5.599 I0423 18:37:15.158011 135932142696256 metric_logger.py:185] completed step: 17, seconds: 18.501, TFLOP/s/device: 0.734, Tokens/s/device: 110.696, total_weights: 65536, loss: 5.511 I0423 18:37:15.305578 135932142696256 metric_logger.py:185] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.651, Tokens/s/device: 13362.477, total_weights: 65536, loss: 5.442 I0423 18:37:15.454087 135932142696256 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0423 18:37:15.454744 135932142696256 checkpointing.py:798] Waited 0.000667572021484375 seconds for step 19 to finish before starting checkpointing. I0423 18:37:15.456737 135932142696256 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:37:19.493365 135805926373120 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_6 I0423 18:37:56.322263 135806983329536 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 26.203 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 60.283798933029175 s) (per-host) I0423 18:37:56.322376 135806983329536 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 59.985149s. I0423 18:38:03.167350 135806983329536 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 66.830109s. I0423 18:38:03.167657 135810716247808 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:38:03.167785 135810716247808 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:38:03.167846 135810716247808 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:38:03.169464 135810716247808 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0423 18:38:03.169696 135932142696256 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. W0423 18:38:03.169826 135932142696256 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.713099 seconds. If this number is high, consider checkpointing less frequently. I0423 18:38:03.171964 135932142696256 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0423 18:38:03.175440 135932142696256 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/19. I0423 18:38:04.144386 135932142696256 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:38:04.144488 135932142696256 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 I0423 18:38:04.188587 135932142696256 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046479s I0423 18:38:04.188750 135932142696256 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.081 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3035733699798584 s) (per-host) I0423 18:38:04.188799 135932142696256 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.303632s (batch_requests_ready=0.252596s, total_serialization_initiated=0.050972s, others=0.000064s) I0423 18:38:04.188894 135932142696256 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.307709s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.307698) I0423 18:38:04.189762 135810707855104 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:38:04.189916 135932142696256 async_checkpointer.py:561] Finished blocking save. Time taken: 1.017886s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/19. I0423 18:38:04.894401 135932142696256 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:38:04.894794 135810716247808 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:38:04.894963 135932142696256 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969435.4567106, 'wait_for_prev_duration_secs': 47.713099241256714, 'time_between_consecutive_saves_sec': 19.799784183502197, 'checkpointer_blocking_start_time': 1776969483.1720033, 'checkpointer_blocking_duration_secs': 1.018019676208496, 'get_old_steps_start_time': 1776969484.190035, 'get_old_steps_duration_secs': 2.0742416381835938e-05, 'checkpoint_manager_blocking_start_time': 1776969435.4549675, 'checkpoint_manager_blocking_duration_secs': 49.439961194992065} I0423 18:38:04.895108 135932142696256 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0423 18:38:04.895152 135932142696256 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:38:09.719881 135811286689536 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_6 I0423 18:38:47.298706 135810707855104 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 36.385 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.41349148750305 s) (per-host) I0423 18:38:47.298827 135810707855104 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.108953s. I0423 18:38:53.446971 135810707855104 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 49.257083s. I0423 18:38:53.447266 135810716247808 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:38:53.447401 135810716247808 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:38:53.447445 135810716247808 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:38:53.448900 135810716247808 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 18:38:53.449061 135932142696256 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. I0423 18:38:53.449252 135932142696256 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 18:38:53.449871 135932142696256 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.516, Tokens/s/device: 13944.969, 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: Thu Apr 23 18:39:04 UTC 2026 EXIT_CODE=0