XPK Start: Sat Apr 18 14:08:33 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 14:08:57.271842: 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 14:08:57.803646 137534970804032 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 14:09:06.844371 137534970804032 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 14:09:06.846754 137534970804032 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-z1r0i-slice-job-0-0.mt-15-ocdbt-true-sv-z1r0i:8482 I0418 14:09:07.818509 137534970804032 max_utils.py:284] Jax distributed system initialized! I0418 14:09:13.063153 137534970804032 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 14:09:13.063255 137534970804032 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 14:09:13.063296 137534970804032 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 14:09:13.063330 137534970804032 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 14:09:13.754926 137534970804032 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 14:09:13.755202 137534970804032 checkpointing.py:677] Setting up checkpoint logger... I0418 14:09:13.755258 137534970804032 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 14:09:13.755303 137534970804032 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 14:09:13.755643 137534970804032 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 0x7d15d5b8fe60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 14:09:17.363836 137534970804032 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 14:09:17.364078 137534970804032 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d02002cb7a0>}, handler_registry=None I0418 14:09:17.364317 137534970804032 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d02002cb7a0>` 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 14:09:17.364365 137534970804032 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d02002ca120>` 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 14:09:17.364402 137534970804032 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 0x7d02002cb7a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d02002cb7a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d02002ca120>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d02002ca120>}). I0418 14:09:17.364978 137534970804032 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 14:09:17.365097 137534970804032 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d01107cbc40> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 14:09:18.238834 137534970804032 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints I0418 14:09:18.925218 137534970804032 checkpoint_manager.py:929] [process=5][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d02002cad50> I0418 14:09:18.925395 137534970804032 checkpointing.py:301] Checkpoint manager created! I0418 14:09:19.847006 137534970804032 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 14:09:19.847117 137534970804032 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:09:20.230000 137534970804032 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:09:20.230088 137534970804032 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:09:20.246309 137534970804032 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:09:20.246369 137534970804032 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:09:20.270474 137534970804032 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 14:09:20.270538 137534970804032 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:09:20.286592 137534970804032 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 14:09:20.286649 137534970804032 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:09:20.302720 137534970804032 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 14:09:20.302779 137534970804032 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:09:20.327195 137534970804032 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 14:09:20.327264 137534970804032 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:09:20.348149 137534970804032 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 14:09:20.348212 137534970804032 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 14:09:20.566082 137534970804032 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 14:09:20.566190 137534970804032 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 14:09:21.983179 137534970804032 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 14:09:21.983305 137534970804032 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 14:09:21.983356 137534970804032 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 14:09:21.983411 137534970804032 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 14:09:21.983476 137534970804032 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 14:09:21.983518 137534970804032 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 14:09:21.983571 137534970804032 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 14:09:21.983621 137534970804032 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 14:09:21.983661 137534970804032 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 14:09:21.983698 137534970804032 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 14:09:21.983745 137534970804032 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 14:09:21.983793 137534970804032 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 14:09:22.470744 137534970804032 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:09:22.470836 137534970804032 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:09:22.485962 137534970804032 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:09:22.486024 137534970804032 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:09:33.519598 137534970804032 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 14:09:33.520440 137534970804032 metric_logger.py:301] number parameters: 1.104 billion I0418 14:09:44.717039 137534970804032 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 14:09:44.871235 137534970804032 checkpointing.py:776] Waited 0.1541764736175537 seconds for step 0 to finish before starting checkpointing. I0418 14:09:44.873841 137534970804032 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:09:44.875645 137534970804032 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 0 I0418 14:09:44.877220 137534970804032 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/0. I0418 14:09:45.605409 137534970804032 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 14:09:45.606428 137534970804032 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:09:45.606502 137534970804032 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 14:09:45.884123 137534970804032 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278788s I0418 14:09:45.884296 137534970804032 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 5.429 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28416013717651367 s) (per-host) I0418 14:09:45.884350 137534970804032 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284224s (batch_requests_ready=0.002151s, total_serialization_initiated=0.282001s, others=0.000072s) I0418 14:09:45.884440 137534970804032 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288540s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.288521) I0418 14:09:45.885263 137534970804032 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/0. I0418 14:09:45.885603 137408821589760 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:29:45.885564 I0418 14:09:45.889794 137534970804032 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:09:45.890040 137405600364288 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:09:45.890165 137534970804032 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776521384.8738232, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776521384.8756824, 'checkpointer_blocking_duration_secs': 1.0100717544555664, 'get_old_steps_start_time': 1776521385.8857791, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776521384.871673, 'checkpoint_manager_blocking_duration_secs': 1.0184528827667236} I0418 14:09:45.890270 137534970804032 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 14:09:45.890319 137534970804032 max_utils.py:750] Memstats: After params initialized: I0418 14:09:45.890366 137534970804032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0)) I0418 14:09:45.890396 137534970804032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0)) I0418 14:09:45.890422 137534970804032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0)) I0418 14:09:45.890445 137534970804032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0)) I0418 14:09:46.201575 137534970804032 metric_logger.py:196] completed step: 0, seconds: 11.196, TFLOP/s/device: 1.214, Tokens/s/device: 182.915, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:09:46.387208 137534970804032 metric_logger.py:196] completed step: 1, seconds: 1.483, TFLOP/s/device: 9.162, Tokens/s/device: 1380.987, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:09:46.808866 137534970804032 metric_logger.py:196] completed step: 2, seconds: 0.038, TFLOP/s/device: 353.675, Tokens/s/device: 53309.733, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719 I0418 14:09:46.956243 137534970804032 metric_logger.py:196] completed step: 3, seconds: 0.422, TFLOP/s/device: 32.219, Tokens/s/device: 4856.337, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506 I0418 14:09:46.960789 137534970804032 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0418 14:09:47.250511 137534970804032 checkpointing.py:776] Waited 0.289654016494751 seconds for step 5 to finish before starting checkpointing. I0418 14:09:47.253278 137534970804032 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:09:48.845102 2796 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 14:09:52.568645 137406137235200 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_5 I0418 14:10:20.802033 137408821589760 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 44.873 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 35.201868772506714 s) (per-host) I0418 14:10:20.802142 137408821589760 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.916427s. I0418 14:10:31.876553 137408821589760 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 45.990822s. I0418 14:10:31.876822 137405600364288 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:10:31.876953 137405600364288 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:10:31.877005 137405600364288 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:10:31.878682 137405600364288 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 14:10:31.878855 137534970804032 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 14:10:31.878986 137534970804032 checkpoint_manager.py:1452] Waiting for previous save to complete took 44.625707 seconds. If this number is high, consider checkpointing less frequently. I0418 14:10:31.881113 137534970804032 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 5 I0418 14:10:31.883592 137534970804032 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/5. I0418 14:10:32.606914 137534970804032 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:10:32.607008 137534970804032 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 14:10:32.647238 137534970804032 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041485s I0418 14:10:32.647399 137534970804032 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 34.386 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0448610782623291 s) (per-host) I0418 14:10:32.647450 137534970804032 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044922s (batch_requests_ready=0.001722s, total_serialization_initiated=0.043131s, others=0.000070s) I0418 14:10:32.647566 137534970804032 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049853s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.049836) I0418 14:10:32.648282 137534970804032 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/5. I0418 14:10:32.648584 137405600364288 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:30:32.648551 I0418 14:10:32.651384 137534970804032 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:10:32.651704 137386600163072 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:10:32.651868 137534970804032 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776521387.2532492, 'wait_for_prev_duration_secs': 44.62570667266846, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776521431.8811526, 'checkpointer_blocking_duration_secs': 0.7675788402557373, 'get_old_steps_start_time': 1776521432.6487558, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1776521387.251193, 'checkpoint_manager_blocking_duration_secs': 45.400639295578} I0418 14:10:32.651988 137534970804032 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0418 14:10:32.652783 137534970804032 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.767, Tokens/s/device: 13379.937, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355 I0418 14:10:32.658796 137534970804032 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.154, Tokens/s/device: 13890.396, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467 I0418 14:10:40.319020 137406137235200 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_5 I0418 14:10:55.858555 137534970804032 metric_logger.py:196] completed step: 6, seconds: 45.697, TFLOP/s/device: 0.297, Tokens/s/device: 44.817, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740 I0418 14:10:56.005888 137534970804032 metric_logger.py:196] completed step: 7, seconds: 23.054, TFLOP/s/device: 0.589, Tokens/s/device: 88.836, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151 I0418 14:10:56.153255 137534970804032 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.379, Tokens/s/device: 13472.177, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098 I0418 14:10:56.300201 137534970804032 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 14:10:56.300883 137534970804032 checkpointing.py:776] Waited 0.0007026195526123047 seconds for step 9 to finish before starting checkpointing. I0418 14:10:56.303131 137534970804032 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:11:17.728825 137405600364288 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 35.004 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.12625789642334 s) (per-host) I0418 14:11:17.728929 137405600364288 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.080232s. I0418 14:11:25.867807 137405600364288 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 53.219094s. I0418 14:11:25.868094 137386600163072 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:11:25.868219 137386600163072 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:11:25.868268 137386600163072 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:11:25.870350 137386600163072 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0418 14:11:25.870552 137534970804032 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0418 14:11:25.870680 137534970804032 checkpoint_manager.py:1452] Waiting for previous save to complete took 29.567550 seconds. If this number is high, consider checkpointing less frequently. I0418 14:11:25.871929 137534970804032 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 9 I0418 14:11:25.874018 137534970804032 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/9. I0418 14:11:27.001671 137534970804032 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:11:27.001767 137534970804032 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 14:11:27.028418 137534970804032 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027665s I0418 14:11:27.028604 137534970804032 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 49.611 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031093120574951172 s) (per-host) I0418 14:11:27.028655 137534970804032 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031154s (batch_requests_ready=0.001618s, total_serialization_initiated=0.029470s, others=0.000066s) I0418 14:11:27.028744 137534970804032 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035179s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.035167) I0418 14:11:27.029445 137534970804032 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.16 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/9. I0418 14:11:27.029759 137386600163072 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:31:27.029726 I0418 14:11:27.037739 137534970804032 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:11:27.038047 137406137235200 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:11:27.038204 137534970804032 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776521456.3031015, 'wait_for_prev_duration_secs': 29.56755018234253, 'time_between_consecutive_saves_sec': 24.42438054084778, 'checkpointer_blocking_start_time': 1776521485.8719676, 'checkpointer_blocking_duration_secs': 1.157930612564087, 'get_old_steps_start_time': 1776521487.0299168, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776521456.30112, 'checkpoint_manager_blocking_duration_secs': 30.737051248550415} I0418 14:11:27.038319 137534970804032 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 14:11:27.038360 137534970804032 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:11:34.352604 137406640535296 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_5 I0418 14:12:10.390634 137386600163072 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 36.402 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.39308524131775 s) (per-host) I0418 14:12:10.390749 137386600163072 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.360884s. I0418 14:12:19.878349 137386600163072 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 52.848469s. I0418 14:12:19.878644 137406137235200 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:12:19.878779 137406137235200 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:12:19.878821 137406137235200 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:12:19.881322 137406137235200 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 14:12:19.881493 137534970804032 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 14:12:19.881646 137534970804032 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:12:19.882613 137534970804032 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.120, Tokens/s/device: 13885.311, 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 14:12:28 UTC 2026 EXIT_CODE=0