XPK Start: Thu Apr 23 17:39:31 UTC 2026 2026-04-23 17:39:35.843896: 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 00:00:1776965975.856838 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776965975.860492 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776965975.871728 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965975.871747 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965975.871749 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965975.871751 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 17:39:55.010336: 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 17:39:55.533534 139054686582592 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 17:40:04,574:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 17:40:04.574588 139054686582592 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 17:40:04,576:jax._src.distributed:157: Connecting to JAX distributed service on mt-04-int8-j7r7u-slice-job-0-0.mt-04-int8-j7r7u:8482 I0423 17:40:04.576984 139054686582592 distributed.py:157] Connecting to JAX distributed service on mt-04-int8-j7r7u-slice-job-0-0.mt-04-int8-j7r7u:8482 I0423 17:40:05.432550 139054686582592 max_utils.py:284] Jax distributed system initialized! I0423 17:40:12.420064 139054686582592 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 17:40:12.420174 139054686582592 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 17:40:12.420219 139054686582592 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 17:40:12.420253 139054686582592 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 17:40:13.038169 139054686582592 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 17:40:13.596870 139054686582592 checkpointing.py:688] Setting up checkpoint logger... I0423 17:40:13.596992 139054686582592 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 17:40:13.597039 139054686582592 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 17:40:13.597253 139054686582592 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 0x7e77bbd79130>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 17:40:16.493852 139054686582592 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 17:40:16.494206 139054686582592 checkpoint_manager.py:709] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6348393260>}, handler_registry=None I0423 17:40:16.494443 139054686582592 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6348393260>` 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 17:40:16.494493 139054686582592 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e63106c4170>` 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 17:40:16.494527 139054686582592 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 0x7e6348393260>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6348393260>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e63106c4170>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e63106c4170>}). I0423 17:40:16.494898 139054686582592 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 17:40:16.494986 139054686582592 async_checkpointer.py:177] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e631043f060> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 17:40:17.672385 139054686582592 checkpoint_manager.py:1818] Found 0 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_04_int8/checkpoints I0423 17:40:18.054093 139054686582592 checkpoint_manager.py:929] [process=5][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=10), 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_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e63106c6570> I0423 17:40:18.054265 139054686582592 checkpointing.py:302] Checkpoint manager created! I0423 17:40:19.508901 139054686582592 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 17:40:19.509027 139054686582592 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0423 17:40:21.890694 139054686582592 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 17:40:21.890817 139054686582592 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:40:21.890873 139054686582592 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:40:21.890917 139054686582592 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 17:40:21.890955 139054686582592 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:40:21.890990 139054686582592 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:40:21.891026 139054686582592 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:40:21.891062 139054686582592 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 17:40:21.891098 139054686582592 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:40:21.891132 139054686582592 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:40:21.891166 139054686582592 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 17:40:21.891201 139054686582592 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 17:40:22.293232 139054686582592 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:40:22.293395 139054686582592 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:40:22.298871 139054686582592 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:40:22.298966 139054686582592 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:40:22.460911 139054686582592 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:40:22.461063 139054686582592 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:40:22.624355 139054686582592 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:40:22.624488 139054686582592 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:40:23.233035 139054686582592 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 17:40:23.233179 139054686582592 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:40:23.396393 139054686582592 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 17:40:23.396529 139054686582592 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:40:23.557528 139054686582592 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 17:40:23.557675 139054686582592 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:40:23.726700 139054686582592 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 17:40:23.726838 139054686582592 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:40:23.836241 139054686582592 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 17:40:23.836371 139054686582592 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 17:40:37.499758 139054686582592 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0423 17:40:37.503387 139054686582592 metric_logger.py:289] number parameters: 1.104 billion I0423 17:40:51.852831 139054686582592 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0423 17:40:52.089902 139054686582592 checkpointing.py:798] Waited 0.2370600700378418 seconds for step 0 to finish before starting checkpointing. I0423 17:40:52.092756 139054686582592 checkpoint_manager.py:2013] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:40:52.094648 139054686582592 checkpoint_manager.py:1518] [process=5] Saving checkpoint at step 0 I0423 17:40:52.097514 139054686582592 async_checkpointer.py:452] [process=5] 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_04_int8/checkpoints/0. I0423 17:40:52.652334 139054686582592 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 17:40:52.653467 139054686582592 jax_array_handlers.py:358] Scheduling D2H of 111 prioritized jax.Array. I0423 17:40:52.653529 139054686582592 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 17:40:52.905105 139054686582592 base_pytree_checkpoint_handler.py:153] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.255222s I0423 17:40:52.905269 139054686582592 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.972 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5190706253051758 s) (per-host) I0423 17:40:52.905326 139054686582592 base_pytree_checkpoint_handler.py:737] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.519138s (batch_requests_ready=0.253845s, total_serialization_initiated=0.265218s, others=0.000075s) I0423 17:40:52.905427 139054686582592 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.523215s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.523196) I0423 17:40:52.906414 138924686436096 async_checkpointer.py:79] [process=5][thread=async_save] Background save thread started. I0423 17:40:52.906590 139054686582592 async_checkpointer.py:561] Finished blocking save. Time taken: 0.811849s. 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_04_int8/checkpoints/0. I0423 17:40:52.919413 139054686582592 checkpoint_manager.py:1566] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:40:52.919676 138922587842304 async_checkpointer.py:265] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:40:52.919807 139054686582592 standard_logger.py:34] {'step': 0, '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_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776966052.0927365, 'wait_for_prev_duration_secs': 6.628036499023438e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776966052.0947056, 'checkpointer_blocking_duration_secs': 0.8120100498199463, 'get_old_steps_start_time': 1776966052.9067352, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776966052.0903716, 'checkpoint_manager_blocking_duration_secs': 0.8293991088867188} I0423 17:40:52.919926 139054686582592 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0423 17:40:52.920010 139054686582592 max_utils.py:750] Memstats: After params initialized: I0423 17:40:52.920060 139054686582592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0)) I0423 17:40:52.920094 139054686582592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0)) I0423 17:40:52.920123 139054686582592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0)) I0423 17:40:52.920149 139054686582592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0)) I0423 17:40:53.192085 139054686582592 metric_logger.py:185] completed step: 0, seconds: 14.348, TFLOP/s/device: 0.947, Tokens/s/device: 142.742, total_weights: 65536, loss: 10.872 I0423 17:40:53.437469 139054686582592 metric_logger.py:185] completed step: 1, seconds: 1.337, TFLOP/s/device: 10.163, Tokens/s/device: 1531.845, total_weights: 65536, loss: 10.872 I0423 17:40:53.875432 139054686582592 metric_logger.py:185] completed step: 2, seconds: 0.020, TFLOP/s/device: 691.070, Tokens/s/device: 104165.607, total_weights: 65536, loss: 10.258 I0423 17:40:54.104617 139054686582592 metric_logger.py:185] completed step: 3, seconds: 0.420, TFLOP/s/device: 32.343, Tokens/s/device: 4875.030, total_weights: 65536, loss: 9.727 I0423 17:40:54.563426 139054686582592 metric_logger.py:185] completed step: 4, seconds: 0.253, TFLOP/s/device: 53.675, Tokens/s/device: 8090.481, total_weights: 65536, loss: 9.270 I0423 17:40:54.572033 139054686582592 metric_logger.py:185] completed step: 5, seconds: 0.229, TFLOP/s/device: 59.337, Tokens/s/device: 8943.895, total_weights: 65536, loss: 8.884 I0423 17:41:16.023881 139054686582592 metric_logger.py:185] completed step: 6, seconds: 0.459, TFLOP/s/device: 29.600, Tokens/s/device: 4461.611, total_weights: 65536, loss: 8.584 I0423 17:41:16.252713 139054686582592 metric_logger.py:185] completed step: 7, seconds: 21.225, TFLOP/s/device: 0.640, Tokens/s/device: 96.489, total_weights: 65536, loss: 8.376 I0423 17:41:16.481790 139054686582592 metric_logger.py:185] completed step: 8, seconds: 0.235, TFLOP/s/device: 57.824, Tokens/s/device: 8715.932, total_weights: 65536, loss: 8.247 I0423 17:41:16.712676 139054686582592 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0423 17:41:16.713562 139054686582592 checkpointing.py:798] Waited 0.0009109973907470703 seconds for step 9 to finish before starting checkpointing. I0423 17:41:16.715689 139054686582592 checkpoint_manager.py:2024] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:41:17.371783 2835 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 17:41:20.037991 138924694828800 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 111 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_04_int8/checkpoints/0/items/array_metadatas/process_5 I0423 17:41:47.860125 138924686436096 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/gbytes_per_sec: 28.475 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 55.47388815879822 s) (per-host) I0423 17:41:47.860231 138924686436096 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 54.953705s. I0423 17:41:55.476629 138924686436096 async_checkpointer.py:144] [process=5][thread=async_save] Background save thread done. Time taken: 62.570086s. I0423 17:41:55.476883 138922587842304 async_checkpointer.py:273] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:41:55.476949 138922587842304 async_checkpointer.py:283] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:41:55.477032 138922587842304 checkpoint_manager.py:2133] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:41:55.479395 138922587842304 checkpoint_manager.py:2142] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 17:41:55.479575 139054686582592 checkpoint_manager.py:2036] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0423 17:41:55.479724 139054686582592 checkpoint_manager.py:1458] Waiting for previous save to complete took 38.764055 seconds. If this number is high, consider checkpointing less frequently. I0423 17:41:55.481688 139054686582592 checkpoint_manager.py:1518] [process=5] Saving checkpoint at step 9 I0423 17:41:55.485098 139054686582592 async_checkpointer.py:452] [process=5] 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_04_int8/checkpoints/9. I0423 17:41:56.045602 139054686582592 jax_array_handlers.py:358] Scheduling D2H of 111 prioritized jax.Array. I0423 17:41:56.045717 139054686582592 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 17:41:56.096249 139054686582592 base_pytree_checkpoint_handler.py:153] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054010s I0423 17:41:56.096413 139054686582592 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.860 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.31737446784973145 s) (per-host) I0423 17:41:56.096466 139054686582592 base_pytree_checkpoint_handler.py:737] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.317437s (batch_requests_ready=0.255325s, total_serialization_initiated=0.062043s, others=0.000069s) I0423 17:41:56.096579 139054686582592 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.321557s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.321540) I0423 17:41:56.097534 138925800675072 async_checkpointer.py:79] [process=5][thread=async_save] Background save thread started. I0423 17:41:56.097708 139054686582592 async_checkpointer.py:561] Finished blocking save. Time taken: 0.615947s. 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_04_int8/checkpoints/9. I0423 17:41:56.106133 139054686582592 checkpoint_manager.py:1566] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:41:56.106424 138922587842304 async_checkpointer.py:265] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:41:56.106587 139054686582592 standard_logger.py:34] {'step': 9, '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_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776966076.7156441, 'wait_for_prev_duration_secs': 38.76405453681946, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776966115.481728, 'checkpointer_blocking_duration_secs': 0.6160871982574463, 'get_old_steps_start_time': 1776966116.0978339, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776966076.7138588, 'checkpoint_manager_blocking_duration_secs': 39.392693281173706} I0423 17:41:56.106722 139054686582592 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0423 17:41:56.106769 139054686582592 checkpoint_manager.py:2024] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:42:01.364644 138925809067776 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 111 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_04_int8/checkpoints/9/items/array_metadatas/process_5 I0423 17:42:37.923378 138925800675072 base_pytree_checkpoint_handler.py:129] [process=5] /jax/checkpoint/write/gbytes_per_sec: 37.481 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.14431309700012 s) (per-host) I0423 17:42:37.923481 138925800675072 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.825837s. I0423 17:42:46.219714 138925800675072 async_checkpointer.py:144] [process=5][thread=async_save] Background save thread done. Time taken: 50.122053s. I0423 17:42:46.220030 138922587842304 async_checkpointer.py:273] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:42:46.220149 138922587842304 async_checkpointer.py:283] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:42:46.220198 138922587842304 checkpoint_manager.py:2133] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:42:46.221471 138922587842304 checkpoint_manager.py:2142] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 17:42:46.221639 139054686582592 checkpoint_manager.py:2036] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0423 17:42:46.221824 139054686582592 checkpoint_manager.py:2013] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:42:46.222523 139054686582592 metric_logger.py:185] completed step: 9, seconds: 0.229, TFLOP/s/device: 59.345, Tokens/s/device: 8945.185, total_weights: 65536, loss: 8.171 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 23 17:42:57 UTC 2026 EXIT_CODE=0