XPK Start: Mon Apr 20 21:08:06 UTC 2026 2026-04-20 21:08:10.686908: 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 00:00:1776719290.699728 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776719290.703440 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776719290.714964 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719290.714982 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719290.714985 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719290.714987 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 21:08:29.829628: 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) I0420 21:08:30.364946 137796638046016 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 21:08:39,404:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 21:08:39.404581 137796638046016 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 21:08:39,406:jax._src.distributed:157: Connecting to JAX distributed service on mt-07-eval-72heb-slice-job-0-0.mt-07-eval-72heb:8482 I0420 21:08:39.406912 137796638046016 distributed.py:157] Connecting to JAX distributed service on mt-07-eval-72heb-slice-job-0-0.mt-07-eval-72heb:8482 I0420 21:08:40.784364 137796638046016 max_utils.py:284] Jax distributed system initialized! I0420 21:08:47.554586 137796638046016 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 21:08:47.554695 137796638046016 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 21:08:47.554735 137796638046016 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 I0420 21:08:47.554769 137796638046016 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 21:08:48.175306 137796638046016 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 21:08:48.291465 137796638046016 checkpointing.py:688] Setting up checkpoint logger... I0420 21:08:48.291588 137796638046016 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 21:08:48.291634 137796638046016 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 21:08:48.291835 137796638046016 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 0x7d52c6575580>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 21:08:51.308699 137796638046016 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 21:08:51.308938 137796638046016 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 0x7d51827828a0>}, handler_registry=None I0420 21:08:51.309175 137796638046016 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d51827828a0>` 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`. I0420 21:08:51.309224 137796638046016 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3f038e0170>` 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`. I0420 21:08:51.309261 137796638046016 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 0x7d51827828a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d51827828a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3f038e0170>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d3f038e0170>}). I0420 21:08:51.309613 137796638046016 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 21:08:51.309692 137796638046016 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d3f03850f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 21:08:52.429066 137796638046016 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints I0420 21:08:52.733890 137796638046016 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=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_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d3f038e03b0> I0420 21:08:52.734068 137796638046016 checkpointing.py:302] Checkpoint manager created! I0420 21:08:53.655155 137796638046016 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 21:08:53.655277 137796638046016 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 21:08:55.097493 137796638046016 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 21:08:55.097624 137796638046016 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:08:55.097680 137796638046016 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:08:55.097725 137796638046016 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 21:08:55.097765 137796638046016 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:08:55.097801 137796638046016 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:08:55.097839 137796638046016 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:08:55.097876 137796638046016 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 21:08:55.097911 137796638046016 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:08:55.097944 137796638046016 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:08:55.097977 137796638046016 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 21:08:55.098011 137796638046016 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 21:08:55.498570 137796638046016 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:08:55.498722 137796638046016 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:08:55.504191 137796638046016 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:08:55.504288 137796638046016 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:08:55.662386 137796638046016 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:08:55.662530 137796638046016 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:08:55.818235 137796638046016 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:08:55.818366 137796638046016 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:08:55.983994 137796638046016 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:08:55.984130 137796638046016 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:08:56.135598 137796638046016 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:08:56.135729 137796638046016 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:08:56.286701 137796638046016 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:08:56.286834 137796638046016 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:08:56.450049 137796638046016 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 21:08:56.450189 137796638046016 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:08:56.472639 137796638046016 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 21:08:56.472736 137796638046016 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 21:09:07.548450 137796638046016 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. I0420 21:09:07.550881 137796638046016 metric_logger.py:289] number parameters: 1.104 billion I0420 21:09:18.729983 137796638046016 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 21:09:18.880903 137796638046016 checkpointing.py:798] Waited 0.1509079933166504 seconds for step 0 to finish before starting checkpointing. I0420 21:09:18.883171 137796638046016 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:09:18.885017 137796638046016 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0420 21:09:18.888296 137796638046016 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/0. I0420 21:09:19.480341 137796638046016 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 21:09:19.481406 137796638046016 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0420 21:09:19.481478 137796638046016 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 I0420 21:09:19.733032 137796638046016 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.253834s I0420 21:09:19.733201 137796638046016 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.981 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5173962116241455 s) (per-host) I0420 21:09:19.733251 137796638046016 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.517455s (batch_requests_ready=0.257002s, total_serialization_initiated=0.260386s, others=0.000066s) I0420 21:09:19.733348 137796638046016 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.521577s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=0.521558) I0420 21:09:19.734288 137670665389824 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0420 21:09:19.734417 137796638046016 async_checkpointer.py:561] Finished blocking save. Time taken: 0.849287s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/0. I0420 21:09:19.780834 137796638046016 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:09:19.781147 137669564671744 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:09:19.781294 137796638046016 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776719358.883154, 'wait_for_prev_duration_secs': 6.341934204101562e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776719358.8851001, 'checkpointer_blocking_duration_secs': 0.8495128154754639, 'get_old_steps_start_time': 1776719359.7346373, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776719358.881332, 'checkpoint_manager_blocking_duration_secs': 0.8999233245849609} I0420 21:09:19.781407 137796638046016 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 21:09:19.781492 137796638046016 max_utils.py:750] Memstats: After params initialized: I0420 21:09:19.781549 137796638046016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0420 21:09:19.781585 137796638046016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0420 21:09:19.781612 137796638046016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0420 21:09:19.781644 137796638046016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0420 21:09:20.047523 137796638046016 metric_logger.py:185] completed step: 0, seconds: 11.178, TFLOP/s/device: 1.216, Tokens/s/device: 183.216, total_weights: 65536, loss: 10.874 I0420 21:09:20.229667 137796638046016 metric_logger.py:185] completed step: 1, seconds: 1.316, TFLOP/s/device: 10.326, Tokens/s/device: 1556.447, total_weights: 65536, loss: 10.874 I0420 21:09:20.650076 137796638046016 metric_logger.py:185] completed step: 2, seconds: 0.037, TFLOP/s/device: 371.700, Tokens/s/device: 56026.700, total_weights: 65536, loss: 10.262 I0420 21:09:20.797518 137796638046016 metric_logger.py:185] completed step: 3, seconds: 0.411, TFLOP/s/device: 33.076, Tokens/s/device: 4985.576, total_weights: 65536, loss: 9.732 I0420 21:09:21.093109 137796638046016 metric_logger.py:185] completed step: 4, seconds: 0.163, TFLOP/s/device: 83.380, Tokens/s/device: 12567.964, total_weights: 65536, loss: 9.274 I0420 21:09:21.099972 137796638046016 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.462, Tokens/s/device: 13936.808, total_weights: 65536, loss: 8.888 I0420 21:09:39.834609 137796638046016 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.900, Tokens/s/device: 6918.592, total_weights: 65536, loss: 8.589 I0420 21:09:39.982253 137796638046016 metric_logger.py:185] completed step: 7, seconds: 18.589, TFLOP/s/device: 0.731, Tokens/s/device: 110.175, total_weights: 65536, loss: 8.380 I0420 21:09:40.129670 137796638046016 metric_logger.py:185] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.816, Tokens/s/device: 13387.283, total_weights: 65536, loss: 8.251 I0420 21:09:40.278090 137796638046016 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 21:09:40.278673 137796638046016 checkpointing.py:798] Waited 0.0005972385406494141 seconds for step 9 to finish before starting checkpointing. I0420 21:09:40.280791 137796638046016 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:09:41.002423 2655 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 21:09:42.884720 137670673782528 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_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/0/items/array_metadatas/process_6 I0420 21:10:14.683586 137670665389824 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 28.478 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 55.46773314476013 s) (per-host) I0420 21:10:14.683719 137670665389824 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 54.949210s. I0420 21:10:22.679762 137670665389824 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 62.945237s. I0420 21:10:22.680062 137669564671744 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:10:22.680184 137669564671744 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:10:22.680235 137669564671744 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:10:22.681782 137669564671744 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 21:10:22.681975 137796638046016 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 21:10:22.682113 137796638046016 checkpoint_manager.py:1458] Waiting for previous save to complete took 42.401325 seconds. If this number is high, consider checkpointing less frequently. I0420 21:10:22.684193 137796638046016 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0420 21:10:22.687584 137796638046016 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/9. I0420 21:10:23.423147 137796638046016 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0420 21:10:23.423306 137796638046016 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 I0420 21:10:23.470538 137796638046016 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.049624s I0420 21:10:23.470705 137796638046016 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.034 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3064563274383545 s) (per-host) I0420 21:10:23.470754 137796638046016 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.306515s (batch_requests_ready=0.252087s, total_serialization_initiated=0.054362s, others=0.000066s) I0420 21:10:23.470846 137796638046016 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.310900s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.310885) I0420 21:10:23.471704 137669588739840 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0420 21:10:23.471860 137796638046016 async_checkpointer.py:561] Finished blocking save. Time taken: 0.787593s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/9. I0420 21:10:23.493765 137796638046016 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:10:23.494074 137669564671744 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:10:23.494240 137796638046016 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776719380.2807631, 'wait_for_prev_duration_secs': 42.40132546424866, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776719422.6842349, 'checkpointer_blocking_duration_secs': 0.7877357006072998, 'get_old_steps_start_time': 1776719423.471989, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1776719380.2789018, 'checkpoint_manager_blocking_duration_secs': 43.21530342102051} I0420 21:10:23.494357 137796638046016 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 21:10:23.494403 137796638046016 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:10:28.665918 137670673782528 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_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_07_eval/checkpoints/9/items/array_metadatas/process_6 I0420 21:11:05.489453 137669588739840 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 37.321 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.325146198272705 s) (per-host) I0420 21:11:05.489558 137669588739840 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.017747s. I0420 21:11:14.366463 137669588739840 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 50.894618s. I0420 21:11:14.366757 137669564671744 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:11:14.366873 137669564671744 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:11:14.366988 137669564671744 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:11:14.368497 137669564671744 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 21:11:14.368684 137796638046016 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 21:11:14.368846 137796638046016 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:11:14.369500 137796638046016 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.255, Tokens/s/device: 13905.675, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 21:11:27 UTC 2026 EXIT_CODE=0