XPK Start: Fri Apr 17 13:46:50 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 13:47:14.185131: 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) I0417 13:47:14.366112 137957407958848 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 13:47:23.405773 137957407958848 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 13:47:23.408300 137957407958848 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-i634k-slice-job-0-0.mt-09-pdb-lt-1-i634k:8482 I0417 13:47:24.763013 137957407958848 max_utils.py:284] Jax distributed system initialized! I0417 13:47:30.725791 137957407958848 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:47:30.725913 137957407958848 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:47:30.725953 137957407958848 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 I0417 13:47:30.725987 137957407958848 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:47:31.749114 137957407958848 maxtext_utils.py:1548] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0417 13:47:31.749387 137957407958848 checkpointing.py:677] Setting up checkpoint logger... I0417 13:47:31.749438 137957407958848 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:47:31.749483 137957407958848 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 13:47:31.749806 137957407958848 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 0x7d77fe84dc10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:47:37.346354 137957407958848 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:47:37.346592 137957407958848 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 0x7d76f2559160>}, handler_registry=None I0417 13:47:37.346830 137957407958848 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d76f2559160>` 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`. I0417 13:47:37.346892 137957407958848 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d63c0493e90>` 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`. I0417 13:47:37.346928 137957407958848 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 0x7d76f2559160>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d76f2559160>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d63c0493e90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d63c0493e90>}). I0417 13:47:37.347253 137957407958848 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 13:47:37.347323 137957407958848 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d62f46c0220> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 13:47:39.105534 137957407958848 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints I0417 13:47:39.778174 137957407958848 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=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/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d63f80a3c80> I0417 13:47:39.778341 137957407958848 checkpointing.py:301] Checkpoint manager created! I0417 13:47:40.686176 137957407958848 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:47:40.686293 137957407958848 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 13:47:41.065911 137957407958848 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:47:41.065999 137957407958848 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 13:47:41.082406 137957407958848 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:47:41.082466 137957407958848 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 13:47:41.106066 137957407958848 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:47:41.106137 137957407958848 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 13:47:41.122327 137957407958848 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:47:41.122387 137957407958848 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 13:47:41.138644 137957407958848 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:47:41.138705 137957407958848 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 13:47:41.163096 137957407958848 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:47:41.163173 137957407958848 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 13:47:41.184813 137957407958848 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:47:41.184888 137957407958848 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0417 13:47:41.399808 137957407958848 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:47:41.399929 137957407958848 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0417 13:47:42.811877 137957407958848 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: ('tensor',) I0417 13:47:42.812006 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, 'tensor') I0417 13:47:42.812060 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, 'tensor') I0417 13:47:42.812118 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: ('tensor', None, 'fsdp') I0417 13:47:42.812170 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: ('tensor', None) I0417 13:47:42.812209 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: ('tensor', None) I0417 13:47:42.812262 137957407958848 maxtext_utils.py:1651] 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, 'tensor', None) I0417 13:47:42.812324 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: ('tensor', None, None, 'fsdp') I0417 13:47:42.812365 137957407958848 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, 'tensor', None) I0417 13:47:42.812402 137957407958848 maxtext_utils.py:1651] 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, 'tensor', None) I0417 13:47:42.812448 137957407958848 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', 'tensor') I0417 13:47:42.812496 137957407958848 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: ('tensor', 'fsdp') I0417 13:47:43.307422 137957407958848 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:47:43.307518 137957407958848 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:47:43.322908 137957407958848 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:47:43.322972 137957407958848 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:47:58.261303 137957407958848 max_utils.py:791] Total memory size: 0.9 GB, Output size: 0.4 GB, Temp size: 0.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0417 13:47:58.268520 137957407958848 metric_logger.py:301] number parameters: 1.104 billion I0417 13:47:59.749961 137957407958848 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:48:13.539916 137957407958848 checkpointing.py:776] Waited 13.78990125656128 seconds for step 0 to finish before starting checkpointing. I0417 13:48:13.542284 137957407958848 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:48:13.544000 137957407958848 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 13:48:13.545421 137957407958848 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0. I0417 13:48:14.685548 137957407958848 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 13:48:14.700065 137957407958848 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 13:48:14.700134 137957407958848 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 I0417 13:48:14.963480 137957407958848 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.264123s I0417 13:48:14.964377 137957407958848 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.693 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27095484733581543 s) (per-host) I0417 13:48:14.964458 137957407958848 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.271058s (batch_requests_ready=0.002605s, total_serialization_initiated=0.267632s, others=0.000820s) I0417 13:48:14.964545 137957407958848 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.275003s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.274986) I0417 13:48:14.965358 137957407958848 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.42 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0. I0417 13:48:14.965720 137829711599360 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:08:14.965683 I0417 13:48:14.976732 137957407958848 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:48:14.977008 137829174728448 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:48:14.977145 137957407958848 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433693.542265, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433693.5440388, 'checkpointer_blocking_duration_secs': 1.4218497276306152, 'get_old_steps_start_time': 1776433694.9659204, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776433693.5404599, 'checkpoint_manager_blocking_duration_secs': 1.4366464614868164} I0417 13:48:14.977299 137957407958848 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:48:14.977352 137957407958848 max_utils.py:750] Memstats: After params initialized: I0417 13:48:14.977402 137957407958848 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:48:14.977435 137957407958848 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:48:14.977461 137957407958848 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 13:48:14.977484 137957407958848 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 13:48:15.330966 137957407958848 metric_logger.py:196] completed step: 0, seconds: 1.154, TFLOP/s/device: 2.942, Tokens/s/device: 443.523, total_weights: 16384, loss: 10.862, lm_loss: 10.862, perplexity: 52153.723 I0417 13:48:15.334358 137957407958848 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/tensorboard/' I0417 13:48:15.367594 137829703206656 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0 I0417 13:48:16.156668 137957407958848 metric_logger.py:196] completed step: 1, seconds: 15.579, TFLOP/s/device: 0.218, Tokens/s/device: 32.864, total_weights: 16384, loss: 10.862, lm_loss: 10.862, perplexity: 52153.723 I0417 13:48:16.213661 137957407958848 metric_logger.py:196] completed step: 2, seconds: 0.826, TFLOP/s/device: 4.110, Tokens/s/device: 619.554, total_weights: 16384, loss: 9.763, lm_loss: 9.763, perplexity: 17373.129 I0417 13:48:16.272144 137957407958848 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 369.416, Tokens/s/device: 55682.436, total_weights: 16384, loss: 8.823, lm_loss: 8.823, perplexity: 6791.817 I0417 13:48:20.408075 137829686421248 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/items I0417 13:48:20.977806 137829154113280 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433699730959517, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776433699730959517, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 13:48:23.357650 2526 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 13:48:29.358672 137829191513856 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_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_0 I0417 13:48:34.967532 137957407958848 metric_logger.py:196] completed step: 4, seconds: 0.058, TFLOP/s/device: 58.921, Tokens/s/device: 8881.180, total_weights: 16384, loss: 7.947, lm_loss: 7.947, perplexity: 2828.417 I0417 13:48:34.976381 137957407958848 metric_logger.py:196] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.157, Tokens/s/device: 8766.073, total_weights: 16384, loss: 7.212, lm_loss: 7.212, perplexity: 1355.339 I0417 13:48:35.033464 137957407958848 metric_logger.py:196] completed step: 6, seconds: 18.694, TFLOP/s/device: 0.182, Tokens/s/device: 27.388, total_weights: 16384, loss: 6.652, lm_loss: 6.652, perplexity: 774.659 I0417 13:48:35.091827 137957407958848 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 519.624, Tokens/s/device: 78323.390, total_weights: 16384, loss: 6.269, lm_loss: 6.269, perplexity: 527.696 I0417 13:48:35.150294 137957407958848 metric_logger.py:196] completed step: 8, seconds: 0.059, TFLOP/s/device: 57.229, Tokens/s/device: 8626.209, total_weights: 16384, loss: 6.031, lm_loss: 6.031, perplexity: 416.165 I0417 13:48:35.208012 137957407958848 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:48:35.208721 137957407958848 checkpointing.py:776] Waited 0.0007193088531494141 seconds for step 9 to finish before starting checkpointing. I0417 13:48:35.210770 137957407958848 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:49:00.821531 137829183121152 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.465170s (commit=34.434385s, array_metadata_write=3.030785s) I0417 13:49:00.822564 137829711599360 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.243 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.12911629676819 s) (per-host) I0417 13:49:00.822686 137829711599360 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.856845s. I0417 13:49:02.945365 137829711599360 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433699730959517, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 13:49:04.210189 137829154113280 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433699730959517, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 13:49:05.306866 137829711599360 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:49:05.371425 137829711599360 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000252s. I0417 13:49:05.984036 137829711599360 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.486351s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/items I0417 13:49:05.985719 137829711599360 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0/items I0417 13:49:07.059832 137829711599360 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0 I0417 13:49:09.315222 137829711599360 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0`. I0417 13:49:09.316045 137829711599360 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 55.77 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/0 I0417 13:49:09.317313 137829711599360 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 54.351471s. I0417 13:49:09.317538 137829174728448 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:49:09.317652 137829174728448 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:49:09.317716 137829174728448 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:49:09.319623 137829174728448 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:49:09.319750 137957407958848 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. W0417 13:49:09.319824 137957407958848 checkpoint_manager.py:1452] Waiting for previous save to complete took 34.109059 seconds. If this number is high, consider checkpointing less frequently. I0417 13:49:09.321953 137957407958848 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 13:49:09.324011 137957407958848 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9. I0417 13:49:10.448977 137957407958848 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 13:49:10.449074 137957407958848 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 I0417 13:49:10.475090 137957407958848 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026798s I0417 13:49:10.475952 137957407958848 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 49.137 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03139352798461914 s) (per-host) I0417 13:49:10.476069 137957407958848 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031540s (batch_requests_ready=0.001860s, total_serialization_initiated=0.028862s, others=0.000818s) I0417 13:49:10.476181 137957407958848 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035499s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.035482) I0417 13:49:10.476805 137957407958848 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.15 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9. I0417 13:49:10.477077 137829166335744 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:09:10.477042 I0417 13:49:10.479207 137957407958848 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:49:10.479431 137829174728448 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:49:10.479563 137957407958848 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433715.2107384, 'wait_for_prev_duration_secs': 34.10905885696411, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433749.3219924, 'checkpointer_blocking_duration_secs': 1.155226707458496, 'get_old_steps_start_time': 1776433750.477242, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1776433715.209005, 'checkpoint_manager_blocking_duration_secs': 35.27052307128906} I0417 13:49:10.479670 137957407958848 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:49:10.479713 137957407958848 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:49:11.110947 137829711599360 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9 I0417 13:49:14.276357 137822740670208 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9/items I0417 13:49:21.318011 137829191513856 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_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_0 I0417 13:49:58.080539 137829183121152 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.470887s (commit=38.309393s, array_metadata_write=2.161494s) I0417 13:49:58.081625 137829166335744 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.159 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 47.637052059173584 s) (per-host) I0417 13:49:58.081739 137829166335744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 47.604549s. I0417 13:50:02.046734 137829166335744 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:50:02.506263 137829166335744 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000246s. I0417 13:50:03.125568 137829166335744 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.881357s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9/items I0417 13:50:03.131018 137829166335744 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9/items I0417 13:50:04.641932 137829166335744 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9 I0417 13:50:08.617300 137829166335744 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9`. I0417 13:50:08.618102 137829166335744 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 59.30 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_09_pdb_lt_1/checkpoints/9 I0417 13:50:08.619429 137829166335744 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 58.142236s. I0417 13:50:08.619569 137829174728448 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:50:08.619641 137829174728448 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:50:08.619699 137829174728448 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:50:08.621689 137829174728448 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:50:08.621889 137957407958848 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. I0417 13:50:08.622039 137957407958848 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:50:08.623005 137957407958848 metric_logger.py:196] completed step: 9, seconds: 0.057, TFLOP/s/device: 59.779, Tokens/s/device: 9010.594, total_weights: 16384, loss: 5.892, lm_loss: 5.892, perplexity: 361.955 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 13:50:21 UTC 2026 EXIT_CODE=0