XPK Start: Mon Apr 20 10:03:32 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 10:03:56.637671: 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 10:03:56.816361 134553489921856 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 10:04:05.857387 134553489921856 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 10:04:05.859752 134553489921856 distributed.py:172] Connecting to JAX distributed service on mt-06-grad-accum-wh1i2-slice-job-0-0.mt-06-grad-accum-wh1i2:8482 I0420 10:04:07.140487 134553489921856 max_utils.py:284] Jax distributed system initialized! I0420 10:04:13.445737 134553489921856 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 10:04:13.445838 134553489921856 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 10:04:13.445878 134553489921856 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0420 10:04:13.445912 134553489921856 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 10:04:14.170110 134553489921856 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 10:04:14.283419 134553489921856 checkpointing.py:688] Setting up checkpoint logger... I0420 10:04:14.283535 134553489921856 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 10:04:14.283581 134553489921856 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 10:04:14.283773 134553489921856 base_pytree_checkpoint_handler.py:441] 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 0x7a5f78b56840>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 10:04:17.115086 134553489921856 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 10:04:17.115366 134553489921856 checkpoint_manager.py:708] [process=2][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a4a3c469e80>}, handler_registry=None I0420 10:04:17.115609 134553489921856 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a4a3c469e80>` 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 10:04:17.115655 134553489921856 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a4a3c494e00>` 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 10:04:17.115687 134553489921856 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 0x7a4a3c469e80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a4a3c469e80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a4a3c494e00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a4a3c494e00>}). I0420 10:04:17.116041 134553489921856 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 10:04:17.116113 134553489921856 async_checkpointer.py:192] [process=2][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a4a3c7a5b20> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 10:04:19.213663 134553489921856 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints I0420 10:04:19.241423 134553489921856 checkpoint_manager.py:929] [process=2][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_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a4a1865c770> I0420 10:04:19.241555 134553489921856 checkpointing.py:302] Checkpoint manager created! I0420 10:04:20.988751 134553489921856 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 10:04:20.988869 134553489921856 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 10:04:22.766753 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0420 10:04:22.766844 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0420 10:04:22.781809 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0420 10:04:22.781866 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0420 10:04:22.811566 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0420 10:04:22.811636 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0420 10:04:22.826476 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0420 10:04:22.826533 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0420 10:04:22.856083 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0420 10:04:22.856155 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0420 10:04:22.871056 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0420 10:04:22.871112 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0420 10:04:22.915558 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0420 10:04:22.915631 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0420 10:04:22.930505 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0420 10:04:22.930562 134553489921856 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0420 10:04:23.157463 134553489921856 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 10:04:23.157657 134553489921856 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 10:04:23.164295 134553489921856 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 10:04:23.164410 134553489921856 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 10:04:23.181128 134553489921856 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 10:04:23.181267 134553489921856 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 10:04:23.197343 134553489921856 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 10:04:23.197476 134553489921856 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 10:04:23.227597 134553489921856 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 10:04:23.227747 134553489921856 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 10:04:23.243909 134553489921856 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 10:04:23.244056 134553489921856 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 10:04:23.260047 134553489921856 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 10:04:23.260182 134553489921856 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 10:04:23.288420 134553489921856 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 10:04:23.288563 134553489921856 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 10:04:23.310150 134553489921856 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 10:04:23.310237 134553489921856 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 10:04:34.994872 134553489921856 max_utils.py:791] Total memory size: 1.7 GB, Output size: 0.4 GB, Temp size: 1.3 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0420 10:04:34.997298 134553489921856 metric_logger.py:301] number parameters: 1.104 billion I0420 10:04:36.630078 134553489921856 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 10:04:47.576561 134553489921856 checkpointing.py:798] Waited 10.946470260620117 seconds for step 0 to finish before starting checkpointing. I0420 10:04:47.579976 134553489921856 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 10:04:47.581752 134553489921856 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 0 I0420 10:04:47.583102 134553489921856 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/0. I0420 10:04:48.306358 134553489921856 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 10:04:48.307311 134553489921856 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 10:04:48.307370 134553489921856 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 10:04:48.580068 134553489921856 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274491s I0420 10:04:48.580239 134553489921856 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 5.438 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2836642265319824 s) (per-host) I0420 10:04:48.580293 134553489921856 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283730s (batch_requests_ready=0.003174s, total_serialization_initiated=0.280482s, others=0.000075s) I0420 10:04:48.580384 134553489921856 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288005s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.287988) I0420 10:04:48.581185 134553489921856 event_tracking.py:125] [process=2] [async] Finished blocking save in 1.00 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/0. I0420 10:04:48.581526 134424976922368 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 10:24:48.581489 I0420 10:04:48.590358 134553489921856 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 10:04:48.590612 134424473622272 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0420 10:04:48.590738 134553489921856 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776679487.5799584, 'wait_for_prev_duration_secs': 7.390975952148438e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776679487.5817902, 'checkpointer_blocking_duration_secs': 0.9998929500579834, 'get_old_steps_start_time': 1776679488.581707, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776679487.5769465, 'checkpoint_manager_blocking_duration_secs': 1.0137522220611572} I0420 10:04:48.590847 134553489921856 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 10:04:48.590912 134553489921856 max_utils.py:750] Memstats: After params initialized: I0420 10:04:48.590962 134553489921856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_8(process=2,(0,2,0,0)) I0420 10:04:48.590993 134553489921856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_9(process=2,(1,2,0,0)) I0420 10:04:48.591032 134553489921856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_12(process=2,(0,3,0,0)) I0420 10:04:48.591058 134553489921856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_13(process=2,(1,3,0,0)) I0420 10:04:48.905389 134553489921856 metric_logger.py:196] completed step: 0, seconds: 1.632, TFLOP/s/device: 33.309, Tokens/s/device: 5020.697, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0420 10:04:49.531368 134553489921856 metric_logger.py:196] completed step: 1, seconds: 12.273, TFLOP/s/device: 4.428, Tokens/s/device: 667.456, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0420 10:04:50.110549 134553489921856 metric_logger.py:196] completed step: 2, seconds: 0.049, TFLOP/s/device: 1118.558, Tokens/s/device: 168601.301, total_weights: 262144, loss: 10.561, lm_loss: 10.561, perplexity: 38601.980 I0420 10:04:50.689155 134553489921856 metric_logger.py:196] completed step: 3, seconds: 0.585, TFLOP/s/device: 92.860, Tokens/s/device: 13996.911, total_weights: 262144, loss: 10.270, lm_loss: 10.270, perplexity: 28865.557 I0420 10:04:51.847142 134553489921856 metric_logger.py:196] completed step: 4, seconds: 0.579, TFLOP/s/device: 93.807, Tokens/s/device: 14139.570, total_weights: 262144, loss: 10.021, lm_loss: 10.021, perplexity: 22497.775 I0420 10:04:51.854569 134553489921856 metric_logger.py:196] completed step: 5, seconds: 0.578, TFLOP/s/device: 93.984, Tokens/s/device: 14166.344, total_weights: 262144, loss: 9.819, lm_loss: 9.819, perplexity: 18383.324 I0420 10:04:52.361291 2586 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 10:04:54.369960 134424482014976 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/0/items/array_metadatas/process_2 I0420 10:05:26.988268 134424976922368 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 40.825 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.69166398048401 s) (per-host) I0420 10:05:26.988373 134424976922368 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.406733s. I0420 10:05:33.049561 134424976922368 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 44.467905s. I0420 10:05:33.049839 134424473622272 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 10:05:33.049953 134424473622272 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0420 10:05:33.050004 134424473622272 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 10:05:33.051896 134424473622272 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 10:05:37.064813 134553489921856 metric_logger.py:196] completed step: 6, seconds: 1.158, TFLOP/s/device: 46.915, Tokens/s/device: 7071.561, total_weights: 262144, loss: 9.666, lm_loss: 9.666, perplexity: 15775.117 I0420 10:05:37.643554 134553489921856 metric_logger.py:196] completed step: 7, seconds: 44.633, TFLOP/s/device: 1.218, Tokens/s/device: 183.543, total_weights: 262144, loss: 9.560, lm_loss: 9.560, perplexity: 14192.143 I0420 10:05:38.222811 134553489921856 metric_logger.py:196] completed step: 8, seconds: 0.585, TFLOP/s/device: 92.963, Tokens/s/device: 14012.329, total_weights: 262144, loss: 9.495, lm_loss: 9.495, perplexity: 13292.714 I0420 10:05:38.802336 134553489921856 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 10:05:38.802989 134553489921856 checkpointing.py:798] Waited 0.0006656646728515625 seconds for step 9 to finish before starting checkpointing. I0420 10:05:38.805511 134553489921856 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 10:05:38.807185 134553489921856 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 9 I0420 10:05:38.808764 134553489921856 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/9. I0420 10:05:39.142233 134553489921856 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 10:05:39.142320 134553489921856 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 10:05:39.183977 134553489921856 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043236s I0420 10:05:39.184166 134553489921856 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 30.631 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05036020278930664 s) (per-host) I0420 10:05:39.184215 134553489921856 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.050420s (batch_requests_ready=0.002810s, total_serialization_initiated=0.047542s, others=0.000068s) I0420 10:05:39.184307 134553489921856 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054655s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.054639) I0420 10:05:39.184997 134553489921856 event_tracking.py:125] [process=2] [async] Finished blocking save in 0.38 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/9. I0420 10:05:39.185338 134424456836864 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 10:25:39.185300 I0420 10:05:39.187362 134553489921856 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 10:05:39.187668 134424485656320 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0420 10:05:39.187802 134553489921856 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776679538.8054795, 'wait_for_prev_duration_secs': 7.128715515136719e-05, 'time_between_consecutive_saves_sec': 5.753482341766357, 'checkpointer_blocking_start_time': 1776679538.807224, 'checkpointer_blocking_duration_secs': 0.37826967239379883, 'get_old_steps_start_time': 1776679539.185516, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776679538.8032799, 'checkpoint_manager_blocking_duration_secs': 0.384488582611084} I0420 10:05:39.187970 134553489921856 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 10:05:39.188027 134553489921856 checkpoint_manager.py:2020] [process=2][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 10:05:44.290603 134424976922368 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260420_081533_06_grad_accum/checkpoints/9/items/array_metadatas/process_2 I0420 10:06:21.823193 134424456836864 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 37.002 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.68935751914978 s) (per-host) I0420 10:06:21.823304 134424456836864 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.637844s. I0420 10:06:28.226950 134424456836864 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 49.041475s. I0420 10:06:28.227254 134424485656320 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 10:06:28.227380 134424485656320 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0420 10:06:28.227429 134424485656320 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 10:06:28.228837 134424485656320 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 10:06:28.229006 134553489921856 checkpoint_manager.py:2032] [process=2][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 10:06:28.229191 134553489921856 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 10:06:28.230104 134553489921856 metric_logger.py:196] completed step: 9, seconds: 0.579, TFLOP/s/device: 93.823, Tokens/s/device: 14141.986, total_weights: 262144, loss: 9.457, lm_loss: 9.457, perplexity: 12793.173 Per train step: Total TFLOPs: 54.35 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 10:06:37 UTC 2026 EXIT_CODE=0