XPK Start: Sat Apr 18 00:30:58 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-18 00:31:23.329642: 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) I0418 00:31:23.863232 134485865555776 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 00:31:32.903596 134485865555776 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 00:31:32.905990 134485865555776 distributed.py:172] Connecting to JAX distributed service on mt-08-checkpoint-async--gjpss-slice-job-0-0.mt-08-checkpoint-async--gjpss:8482 I0418 00:31:33.985995 134485865555776 max_utils.py:284] Jax distributed system initialized! I0418 00:31:40.197192 134485865555776 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 00:31:40.197295 134485865555776 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 00:31:40.197336 134485865555776 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 I0418 00:31:40.197368 134485865555776 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 00:31:40.895281 134485865555776 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 00:31:40.895553 134485865555776 checkpointing.py:677] Setting up checkpoint logger... I0418 00:31:40.895608 134485865555776 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 00:31:40.895654 134485865555776 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 00:31:40.896014 134485865555776 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 0x7a4fd0314650>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 00:31:46.354021 134485865555776 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 00:31:46.354252 134485865555776 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a3b344183b0>}, handler_registry=None I0418 00:31:46.354483 134485865555776 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a3b344183b0>` 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`. I0418 00:31:46.354531 134485865555776 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a3b34102300>` 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`. I0418 00:31:46.354571 134485865555776 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 0x7a3b344183b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a3b344183b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a3b34102300>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a3b34102300>}). I0418 00:31:46.354922 134485865555776 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 00:31:46.354994 134485865555776 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a3a3c41b9c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 00:31:47.034917 134485865555776 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints I0418 00:31:47.986783 134485865555776 checkpoint_manager.py:929] [process=0][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/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a3b340f62d0> I0418 00:31:47.986994 134485865555776 checkpointing.py:301] Checkpoint manager created! I0418 00:31:48.911250 134485865555776 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 00:31:48.911366 134485865555776 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 00:31:49.286424 134485865555776 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 00:31:49.286513 134485865555776 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 00:31:49.302899 134485865555776 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 00:31:49.302957 134485865555776 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 00:31:49.326603 134485865555776 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 00:31:49.326674 134485865555776 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 00:31:49.342960 134485865555776 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 00:31:49.343015 134485865555776 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 00:31:49.359241 134485865555776 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 00:31:49.359300 134485865555776 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 00:31:49.384763 134485865555776 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 00:31:49.384833 134485865555776 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 00:31:49.405945 134485865555776 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 00:31:49.406011 134485865555776 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 00:31:49.623468 134485865555776 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 00:31:49.623573 134485865555776 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 00:31:51.053525 134485865555776 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 00:31:51.053648 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 00:31:51.053701 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 00:31:51.053757 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0418 00:31:51.053808 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 00:31:51.053858 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 00:31:51.053913 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0418 00:31:51.053967 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0418 00:31:51.054006 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0418 00:31:51.054042 134485865555776 maxtext_utils.py:1654] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0418 00:31:51.054090 134485865555776 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 00:31:51.054136 134485865555776 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 00:31:51.542857 134485865555776 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 00:31:51.542951 134485865555776 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 00:31:51.558295 134485865555776 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 00:31:51.558356 134485865555776 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 00:32:02.614019 134485865555776 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. I0418 00:32:02.621818 134485865555776 metric_logger.py:301] number parameters: 1.104 billion I0418 00:32:14.145167 134485865555776 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 00:32:14.298283 134485865555776 checkpointing.py:776] Waited 0.15310144424438477 seconds for step 0 to finish before starting checkpointing. I0418 00:32:14.300788 134485865555776 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 00:32:14.303159 134485865555776 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0418 00:32:14.304652 134485865555776 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0. I0418 00:32:15.465572 134485865555776 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 00:32:15.480139 134485865555776 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 00:32:15.480206 134485865555776 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 I0418 00:32:15.731092 134355820480256 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0 I0418 00:32:15.749090 134485865555776 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269659s I0418 00:32:15.750043 134485865555776 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.584 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27623748779296875 s) (per-host) I0418 00:32:15.750108 134485865555776 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.276318s (batch_requests_ready=0.002213s, total_serialization_initiated=0.273248s, others=0.000856s) I0418 00:32:15.750185 134485865555776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.280379s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.280360) I0418 00:32:15.750935 134485865555776 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.45 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0. I0418 00:32:15.751229 134355246167808 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 00:52:15.751193 I0418 00:32:15.765290 134485865555776 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 00:32:15.765558 134355828872960 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0418 00:32:15.765676 134485865555776 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776472334.3007636, 'wait_for_prev_duration_secs': 0.00010728836059570312, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776472334.3032148, 'checkpointer_blocking_duration_secs': 1.4481568336486816, 'get_old_steps_start_time': 1776472335.7513947, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776472334.2988272, 'checkpoint_manager_blocking_duration_secs': 1.4668118953704834} I0418 00:32:15.765779 134485865555776 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 00:32:15.765826 134485865555776 max_utils.py:750] Memstats: After params initialized: I0418 00:32:15.765892 134485865555776 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0418 00:32:15.765925 134485865555776 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0418 00:32:15.765953 134485865555776 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0418 00:32:15.765977 134485865555776 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0418 00:32:16.098721 134485865555776 metric_logger.py:196] completed step: 0, seconds: 11.224, TFLOP/s/device: 1.211, Tokens/s/device: 182.461, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 00:32:16.102165 134485865555776 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/tensorboard/' I0418 00:32:16.496586 134485865555776 metric_logger.py:196] completed step: 1, seconds: 1.952, TFLOP/s/device: 6.962, Tokens/s/device: 1049.346, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 00:32:16.642533 134485865555776 metric_logger.py:196] completed step: 2, seconds: 0.399, TFLOP/s/device: 34.086, Tokens/s/device: 5137.841, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719 I0418 00:32:16.789847 134485865555776 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 1447.441, Tokens/s/device: 218174.071, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506 I0418 00:32:16.799284 134485865555776 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0418 00:32:17.084101 134485865555776 checkpointing.py:776] Waited 0.28467607498168945 seconds for step 5 to finish before starting checkpointing. I0418 00:32:17.087070 134485865555776 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 00:32:18.791094 134355803694848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/items I0418 00:32:18.991603 134354198062848 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776472338542414329, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776472338542414329, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0418 00:32:20.791822 2909 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 00:32:24.689517 134355296958208 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/items/array_metadatas/process_0 I0418 00:32:57.049915 134355288565504 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.272857s (commit=35.070560s, array_metadata_write=1.202297s) I0418 00:32:57.051357 134355246167808 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 37.992 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.57750916481018 s) (per-host) I0418 00:32:57.051521 134355246167808 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.300178s. I0418 00:32:58.254254 134355246167808 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776472338542414329, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0418 00:32:59.384702 134355246167808 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000410s. I0418 00:32:59.420367 134354198062848 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776472338542414329, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0418 00:32:59.730123 134355246167808 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0418 00:33:00.638705 134355246167808 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.262342s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/items I0418 00:33:00.640806 134355246167808 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0/items I0418 00:33:01.664927 134355246167808 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0 I0418 00:33:04.372975 134355246167808 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0`. I0418 00:33:04.374109 134355246167808 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.07 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/0 I0418 00:33:04.375593 134355246167808 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.624246s. I0418 00:33:04.375912 134355828872960 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 00:33:04.376074 134355828872960 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0418 00:33:04.376197 134355828872960 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 00:33:04.378984 134355828872960 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 00:33:04.379271 134485865555776 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 00:33:04.379414 134485865555776 checkpoint_manager.py:1452] Waiting for previous save to complete took 47.292342 seconds. If this number is high, consider checkpointing less frequently. I0418 00:33:04.381251 134485865555776 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0418 00:33:04.383637 134485865555776 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5. I0418 00:33:05.524793 134485865555776 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 00:33:05.524958 134485865555776 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 I0418 00:33:05.566283 134485865555776 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042539s I0418 00:33:05.567350 134485865555776 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 30.156 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05115365982055664 s) (per-host) I0418 00:33:05.567461 134485865555776 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.051290s (batch_requests_ready=0.004004s, total_serialization_initiated=0.046324s, others=0.000961s) I0418 00:33:05.567597 134485865555776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.056196s (all_items=0.000025s, per_item={'items': '0.00002456'}, temp_paths=0.056172) I0418 00:33:05.568635 134485865555776 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.19 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5. I0418 00:33:05.569039 134355828872960 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 00:53:05.569000 I0418 00:33:05.571598 134485865555776 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0418 00:33:05.571939 134354755905280 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0418 00:33:05.572197 134485865555776 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776472337.0870223, 'wait_for_prev_duration_secs': 47.29234218597412, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776472384.3813286, 'checkpointer_blocking_duration_secs': 1.187903881072998, 'get_old_steps_start_time': 1776472385.5692647, 'get_old_steps_duration_secs': 4.220008850097656e-05, 'checkpoint_manager_blocking_start_time': 1776472337.0847542, 'checkpoint_manager_blocking_duration_secs': 48.48737359046936} I0418 00:33:05.572372 134485865555776 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0418 00:33:05.755793 134355246167808 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5 I0418 00:33:08.775803 134354739119872 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5/items I0418 00:33:14.213153 134355296958208 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5/items/array_metadatas/process_0 I0418 00:33:30.457569 134485865555776 metric_logger.py:196] completed step: 4, seconds: 0.147, TFLOP/s/device: 92.392, Tokens/s/device: 13926.383, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355 I0418 00:33:30.466911 134485865555776 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.279, Tokens/s/device: 13909.358, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467 I0418 00:33:30.612478 134485865555776 metric_logger.py:196] completed step: 6, seconds: 73.666, TFLOP/s/device: 0.184, Tokens/s/device: 27.801, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740 I0418 00:33:30.760087 134485865555776 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1966.299, Tokens/s/device: 296382.055, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151 I0418 00:33:30.907289 134485865555776 metric_logger.py:196] completed step: 8, seconds: 0.147, TFLOP/s/device: 92.625, Tokens/s/device: 13961.415, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098 I0418 00:33:31.054003 134485865555776 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 00:33:31.054704 134485865555776 checkpointing.py:776] Waited 0.0007228851318359375 seconds for step 9 to finish before starting checkpointing. I0418 00:33:31.056871 134485865555776 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 00:33:53.604575 134355288565504 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.830868s (commit=39.858637s, array_metadata_write=2.972231s) I0418 00:33:53.607131 134355828872960 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 32.846 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 48.09091925621033 s) (per-host) I0418 00:33:53.607267 134355828872960 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 48.038084s. I0418 00:33:56.872696 134355828872960 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0418 00:33:56.915744 134355828872960 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000237s. I0418 00:33:57.941179 134355828872960 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.473904s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5/items I0418 00:33:57.944022 134355828872960 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5/items I0418 00:33:58.932578 134355828872960 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5 I0418 00:34:01.119596 134355828872960 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5`. I0418 00:34:01.120474 134355828872960 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 56.74 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/5 I0418 00:34:01.121943 134355828872960 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 55.552758s. I0418 00:34:01.122077 134354755905280 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 00:34:01.122140 134354755905280 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0418 00:34:01.122208 134354755905280 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0418 00:34:01.125103 134354755905280 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0418 00:34:01.125298 134485865555776 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0418 00:34:01.125393 134485865555776 checkpoint_manager.py:1452] Waiting for previous save to complete took 30.068546 seconds. If this number is high, consider checkpointing less frequently. I0418 00:34:01.126753 134485865555776 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0418 00:34:01.128721 134485865555776 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9. I0418 00:34:02.277660 134485865555776 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 00:34:02.277756 134485865555776 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 I0418 00:34:02.304265 134485865555776 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027540s I0418 00:34:02.305042 134485865555776 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 48.288 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031945228576660156 s) (per-host) I0418 00:34:02.305120 134485865555776 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.032042s (batch_requests_ready=0.001862s, total_serialization_initiated=0.029445s, others=0.000735s) I0418 00:34:02.305196 134485865555776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035937s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.035921) I0418 00:34:02.305894 134485865555776 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.18 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9. I0418 00:34:02.306206 134354755905280 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 00:54:02.306169 I0418 00:34:02.314208 134485865555776 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 00:34:02.314486 134355246167808 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0418 00:34:02.314643 134485865555776 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776472411.0568182, 'wait_for_prev_duration_secs': 30.068545818328857, 'time_between_consecutive_saves_sec': 26.677762269973755, 'checkpointer_blocking_start_time': 1776472441.126791, 'checkpointer_blocking_duration_secs': 1.1795520782470703, 'get_old_steps_start_time': 1776472442.3063624, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776472411.0549657, 'checkpoint_manager_blocking_duration_secs': 31.259645462036133} I0418 00:34:02.314759 134485865555776 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 00:34:02.314801 134485865555776 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 00:34:02.957214 134355828872960 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9 I0418 00:34:06.679601 134355820480256 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9/items I0418 00:34:12.099025 134355296958208 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9/items/array_metadatas/process_0 I0418 00:34:51.747230 134355288565504 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 43.086659s (commit=41.050502s, array_metadata_write=2.036157s) I0418 00:34:51.748244 134354755905280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 31.927 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 49.475138902664185 s) (per-host) I0418 00:34:51.748315 134354755905280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.441998s. I0418 00:34:54.263653 134354755905280 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0418 00:34:55.361031 134354755905280 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000253s. I0418 00:34:55.361220 134354755905280 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.937346s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9/items I0418 00:34:55.362553 134354755905280 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9/items I0418 00:34:55.902679 134354755905280 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9 I0418 00:34:58.195453 134354755905280 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9`. I0418 00:34:58.196222 134354755905280 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 57.07 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_235819/linen_xpk_main_20260417_235819_08_checkpoint_async_true/checkpoints/9 I0418 00:34:58.197504 134354755905280 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 55.891185s. I0418 00:34:58.197703 134355246167808 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 00:34:58.197818 134355246167808 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0418 00:34:58.197887 134355246167808 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 00:34:58.199616 134355246167808 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 00:34:58.199795 134485865555776 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 00:34:58.199974 134485865555776 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 00:34:58.200914 134485865555776 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 92.101, Tokens/s/device: 13882.487, total_weights: 65536, loss: 7.181, lm_loss: 7.181, perplexity: 1314.495 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sat Apr 18 00:35:09 UTC 2026 EXIT_CODE=0