XPK Start: Thu Apr 23 18:43:44 UTC 2026 2026-04-23 18:43:48.824981: 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 00:00:1776969828.838088 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776969828.841837 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776969828.853862 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969828.853882 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969828.853884 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776969828.853886 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 18:44:08.156216: 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 18:44:08.688344 134243729827648 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 18:44:17,728:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 18:44:17.728957 134243729827648 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 18:44:17,731:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-false-rs-ur41g-slice-job-0-0.mt-15-ocdbt-false-rs-ur41g:8482 I0423 18:44:17.731379 134243729827648 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-ur41g-slice-job-0-0.mt-15-ocdbt-false-rs-ur41g:8482 I0423 18:44:18.744329 134243729827648 max_utils.py:284] Jax distributed system initialized! I0423 18:44:25.795955 134243729827648 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 18:44:25.796064 134243729827648 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 18:44:25.796121 134243729827648 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 18:44:25.796160 134243729827648 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 18:44:26.421272 134243729827648 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 18:44:26.536011 134243729827648 checkpointing.py:688] Setting up checkpoint logger... I0423 18:44:26.536146 134243729827648 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0423 18:44:26.536193 134243729827648 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 18:44:26.536403 134243729827648 base_pytree_checkpoint_handler.py:415] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7a178bf4c680>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 18:44:27.816825 134243729827648 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 18:44:27.817166 134243729827648 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 0x7a16495696a0>}, handler_registry=None I0423 18:44:27.817410 134243729827648 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a16495696a0>` 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 18:44:27.817460 134243729827648 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02fc5bc590>` 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 18:44:27.817497 134243729827648 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 0x7a16495696a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a16495696a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02fc5bc590>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a02fc5bc590>}). I0423 18:44:27.817826 134243729827648 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 18:44:27.817904 134243729827648 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a02fc4e5260> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 18:44:30.127730 134123070928640 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776969680561867101, 'commit_timestamp_nsecs': 1776969696531324403, '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_false/checkpoints/0/_CHECKPOINT_METADATA I0423 18:44:30.372989 134123070928640 checkpoint_manager.py:1877] Missing metrics for step 0 I0423 18:44:30.645860 134243729827648 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_false/checkpoints I0423 18:44:30.648428 134243729827648 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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a02fc431040> I0423 18:44:30.648554 134243729827648 checkpointing.py:302] Checkpoint manager created! I0423 18:44:31.607850 134243729827648 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 18:44:31.607972 134243729827648 checkpointing.py:582] restoring from this run's directory step 9 I0423 18:44:32.095885 134243729827648 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_false/checkpoints/9. I0423 18:44:35.280336 1457 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 18:44:45.593869 134243729827648 jax_array_handlers.py:837] [process=6] /jax/orbax/read/worker/io/requested throughput: 132.295 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.94512939453125 s) (per-host) I0423 18:44:45.614746 134243729827648 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/read/gbytes_per_sec: 127.824 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 12.381150245666504 s) (per-host) I0423 18:44:46.592611 134243729827648 checkpointer.py:318] Finished restoring checkpoint in 14.74 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_false/checkpoints/9. I0423 18:44:46.592767 134243729827648 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_false/checkpoints', 'checkpointer_start_time': 1776969871.855156, 'checkpointer_duration_secs': 14.737579584121704, 'checkpoint_manager_start_time': 1776969871.608586, 'checkpoint_manager_duration_secs': 14.98415231704712} fsdp: 32 I0423 18:44:46.674382 134243729827648 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 18:44:46.674501 134243729827648 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 18:44:46.674552 134243729827648 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 18:44:46.674601 134243729827648 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 18:44:46.674640 134243729827648 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 18:44:46.674674 134243729827648 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 18:44:46.674712 134243729827648 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:44:46.674749 134243729827648 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 18:44:46.674785 134243729827648 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:44:46.674818 134243729827648 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 18:44:46.674851 134243729827648 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 18:44:46.674884 134243729827648 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 18:44:47.073332 134243729827648 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 18:44:47.073484 134243729827648 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:44:47.078933 134243729827648 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 18:44:47.079030 134243729827648 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:44:47.233962 134243729827648 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 18:44:47.234111 134243729827648 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:44:47.388845 134243729827648 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 18:44:47.388983 134243729827648 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 18:44:47.559293 134243729827648 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 18:44:47.559438 134243729827648 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:44:47.714338 134243729827648 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 18:44:47.714475 134243729827648 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:44:47.866065 134243729827648 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 18:44:47.866233 134243729827648 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:44:48.028773 134243729827648 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 18:44:48.028908 134243729827648 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 18:44:48.051658 134243729827648 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 18:44:48.051754 134243729827648 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 18:44:59.383619 134243729827648 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 18:44:59.385996 134243729827648 metric_logger.py:289] number parameters: 1.104 billion I0423 18:45:10.561538 134243729827648 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0423 18:45:10.712085 134243729827648 checkpointing.py:798] Waited 0.15053701400756836 seconds for step 10 to finish before starting checkpointing. I0423 18:45:10.714339 134243729827648 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 18:45:10.715836 134243729827648 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0423 18:45:10.718484 134243729827648 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_false/checkpoints/10. I0423 18:45:11.297149 134243729827648 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 18:45:11.298091 134243729827648 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:45:11.298162 134243729827648 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 18:45:11.548348 134243729827648 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.252343s I0423 18:45:11.548530 134243729827648 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.025 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.510023832321167 s) (per-host) I0423 18:45:11.548602 134243729827648 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.510108s (batch_requests_ready=0.253046s, total_serialization_initiated=0.256968s, others=0.000094s) I0423 18:45:11.548717 134243729827648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.514293s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.514273) I0423 18:45:11.549670 134120847943424 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:45:11.549799 134243729827648 async_checkpointer.py:561] Finished blocking save. Time taken: 0.833892s. 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_false/checkpoints/10. I0423 18:45:11.579579 134243729827648 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:45:11.579868 134117660288768 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:45:11.580001 134243729827648 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969910.71432, 'wait_for_prev_duration_secs': 6.4849853515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776969910.7158747, 'checkpointer_blocking_duration_secs': 0.8341052532196045, 'get_old_steps_start_time': 1776969911.5500004, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776969910.7124877, 'checkpoint_manager_blocking_duration_secs': 0.8674778938293457} I0423 18:45:11.580123 134243729827648 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0423 18:45:11.580196 134243729827648 max_utils.py:750] Memstats: After params initialized: I0423 18:45:11.580247 134243729827648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 18:45:11.580279 134243729827648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 18:45:11.580309 134243729827648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 18:45:11.580341 134243729827648 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 18:45:11.852683 134243729827648 metric_logger.py:185] completed step: 10, seconds: 11.175, TFLOP/s/device: 1.216, Tokens/s/device: 183.274, total_weights: 65536, loss: 6.827 I0423 18:45:12.025415 134243729827648 metric_logger.py:185] completed step: 11, seconds: 1.289, TFLOP/s/device: 10.537, Tokens/s/device: 1588.272, total_weights: 65536, loss: 6.522 I0423 18:45:12.172662 134243729827648 metric_logger.py:185] completed step: 12, seconds: 0.017, TFLOP/s/device: 785.066, Tokens/s/device: 118333.622, total_weights: 65536, loss: 6.260 I0423 18:45:12.320446 134243729827648 metric_logger.py:185] completed step: 13, seconds: 0.164, TFLOP/s/device: 83.043, Tokens/s/device: 12517.113, total_weights: 65536, loss: 6.039 I0423 18:45:12.327829 134243729827648 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0423 18:45:12.615283 134243729827648 checkpointing.py:798] Waited 0.28742384910583496 seconds for step 15 to finish before starting checkpointing. I0423 18:45:12.617890 134243729827648 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:45:16.755888 134118205552384 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_false/checkpoints/10/items/array_metadatas/process_6 I0423 18:45:22.637512 134120847943424 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 136.184 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.598975896835327 s) (per-host) I0423 18:45:22.637635 134120847943424 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.087754s. I0423 18:45:28.916519 134120847943424 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 17.366622s. I0423 18:45:28.916755 134117660288768 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:45:28.916821 134117660288768 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:45:28.916876 134117660288768 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:45:28.918280 134117660288768 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 18:45:28.918480 134243729827648 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 18:45:28.918627 134243729827648 checkpoint_manager.py:1458] Waiting for previous save to complete took 16.300742 seconds. If this number is high, consider checkpointing less frequently. I0423 18:45:28.920593 134243729827648 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 15 I0423 18:45:28.924737 134243729827648 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_false/checkpoints/15. I0423 18:45:29.461032 134243729827648 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:45:29.461146 134243729827648 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 18:45:29.505420 134243729827648 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046400s I0423 18:45:29.505595 134243729827648 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.076 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3039207458496094 s) (per-host) I0423 18:45:29.505648 134243729827648 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.303991s (batch_requests_ready=0.252516s, total_serialization_initiated=0.051398s, others=0.000077s) I0423 18:45:29.505741 134243729827648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.308881s (all_items=0.000013s, per_item={'items': '0.00001335'}, temp_paths=0.308868) I0423 18:45:29.506604 134118197159680 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:45:29.506766 134243729827648 async_checkpointer.py:561] Finished blocking save. Time taken: 0.586099s. 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_false/checkpoints/15. I0423 18:45:29.523972 134243729827648 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:45:29.524247 134117660288768 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:45:29.524378 134243729827648 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969912.6178608, 'wait_for_prev_duration_secs': 16.30074191093445, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776969928.9206326, 'checkpointer_blocking_duration_secs': 0.5862417221069336, 'get_old_steps_start_time': 1776969929.5068946, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776969912.6156533, 'checkpoint_manager_blocking_duration_secs': 16.908690929412842} I0423 18:45:29.524494 134243729827648 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0423 18:45:29.525279 134243729827648 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.732, Tokens/s/device: 13977.614, total_weights: 65536, loss: 5.858 I0423 18:45:29.532711 134243729827648 metric_logger.py:185] completed step: 15, seconds: 0.149, TFLOP/s/device: 91.134, Tokens/s/device: 13736.669, total_weights: 65536, loss: 5.713 I0423 18:45:47.970443 134243729827648 metric_logger.py:185] completed step: 16, seconds: 17.203, TFLOP/s/device: 0.790, Tokens/s/device: 119.047, total_weights: 65536, loss: 5.599 I0423 18:45:48.117876 134243729827648 metric_logger.py:185] completed step: 17, seconds: 18.292, TFLOP/s/device: 0.743, Tokens/s/device: 111.959, total_weights: 65536, loss: 5.511 I0423 18:45:48.265479 134243729827648 metric_logger.py:185] completed step: 18, seconds: 0.154, TFLOP/s/device: 88.451, Tokens/s/device: 13332.292, total_weights: 65536, loss: 5.442 I0423 18:45:48.413830 134243729827648 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0423 18:45:48.414465 134243729827648 checkpointing.py:798] Waited 0.0006468296051025391 seconds for step 19 to finish before starting checkpointing. I0423 18:45:48.416258 134243729827648 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:45:52.918229 134118205552384 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_false/checkpoints/15/items/array_metadatas/process_6 I0423 18:45:58.522171 134118197159680 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 53.874 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 29.320443630218506 s) (per-host) I0423 18:45:58.522285 134118197159680 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.015572s. I0423 18:46:04.550199 134118197159680 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 35.043470s. I0423 18:46:04.550505 134117660288768 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:46:04.550628 134117660288768 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:46:04.550677 134117660288768 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:46:04.552038 134117660288768 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0423 18:46:04.552220 134243729827648 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 18:46:04.552355 134243729827648 checkpoint_manager.py:1458] Waiting for previous save to complete took 16.136112 seconds. If this number is high, consider checkpointing less frequently. I0423 18:46:04.554010 134243729827648 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0423 18:46:04.557472 134243729827648 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_false/checkpoints/19. I0423 18:46:05.530851 134243729827648 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 18:46:05.530950 134243729827648 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 18:46:05.575863 134243729827648 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.047059s I0423 18:46:05.576027 134243729827648 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.989 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3091726303100586 s) (per-host) I0423 18:46:05.576073 134243729827648 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.309228s (batch_requests_ready=0.257286s, total_serialization_initiated=0.051882s, others=0.000060s) I0423 18:46:05.576180 134243729827648 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.313212s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.313200) I0423 18:46:05.576919 134116049676032 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 18:46:05.577074 134243729827648 async_checkpointer.py:561] Finished blocking save. Time taken: 1.023004s. 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_false/checkpoints/19. I0423 18:46:05.598898 134243729827648 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 18:46:05.599223 134117660288768 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 18:46:05.599381 134243729827648 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776969948.416227, 'wait_for_prev_duration_secs': 16.136112451553345, 'time_between_consecutive_saves_sec': 19.497905015945435, 'checkpointer_blocking_start_time': 1776969964.5540495, 'checkpointer_blocking_duration_secs': 1.0231130123138428, 'get_old_steps_start_time': 1776969965.5771756, 'get_old_steps_duration_secs': 2.0265579223632812e-05, 'checkpoint_manager_blocking_start_time': 1776969948.414684, 'checkpoint_manager_blocking_duration_secs': 17.184666633605957} I0423 18:46:05.599483 134243729827648 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0423 18:46:05.599532 134243729827648 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 18:46:10.673171 134118205552384 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_false/checkpoints/19/items/array_metadatas/process_6 I0423 18:46:16.094801 134116049676032 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 145.882 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.827908277511597 s) (per-host) I0423 18:46:16.094920 134116049676032 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.517885s. I0423 18:46:21.538141 134116049676032 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 15.961091s. I0423 18:46:21.538435 134117660288768 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 18:46:21.538558 134117660288768 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 18:46:21.538618 134117660288768 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 18:46:21.539893 134117660288768 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 18:46:21.540086 134243729827648 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 18:46:21.540293 134243729827648 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 18:46:21.541013 134243729827648 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.370, Tokens/s/device: 13923.070, 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 18:46:35 UTC 2026 EXIT_CODE=0