XPK Start: Sat Apr 18 14:18:45 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 14:19:10.097699: 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 14:19:10.278211 140206216668992 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 14:19:19.318382 140206216668992 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 14:19:19.320711 140206216668992 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-i5n2j-slice-job-0-0.mt-15-ocdbt-false-sv-i5n2j:8482 I0418 14:19:20.425721 140206216668992 max_utils.py:284] Jax distributed system initialized! I0418 14:19:26.482387 140206216668992 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 14:19:26.482490 140206216668992 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 14:19:26.482530 140206216668992 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 14:19:26.482564 140206216668992 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 14:19:27.177981 140206216668992 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 14:19:27.178274 140206216668992 checkpointing.py:677] Setting up checkpoint logger... I0418 14:19:27.178330 140206216668992 checkpointing.py:233] Creating checkpoint manager with ocdbt=False and zarr3=True I0418 14:19:27.178375 140206216668992 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 14:19:27.178712 140206216668992 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7f8396134560>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 14:19:30.534573 140206216668992 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 14:19:30.534809 140206216668992 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f6f2c464c80>}, handler_registry=None I0418 14:19:30.535135 140206216668992 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f6f2c464c80>` 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 14:19:30.535197 140206216668992 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f6f2c30eb10>` 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 14:19:30.535242 140206216668992 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 0x7f6f2c464c80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f6f2c464c80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f6f2c30eb10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f6f2c30eb10>}). I0418 14:19:30.535572 140206216668992 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 14:19:30.535645 140206216668992 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7f6e446bc180> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 14:19:31.705247 140206216668992 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_false/checkpoints I0418 14:19:32.137992 140206216668992 checkpoint_manager.py:929] [process=7][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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f6f2c307b30> I0418 14:19:32.138194 140206216668992 checkpointing.py:301] Checkpoint manager created! I0418 14:19:33.068305 140206216668992 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 14:19:33.068423 140206216668992 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:19:33.446784 140206216668992 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:19:33.446877 140206216668992 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:19:33.463089 140206216668992 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 14:19:33.463148 140206216668992 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 14:19:33.486799 140206216668992 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 14:19:33.486870 140206216668992 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:19:33.503262 140206216668992 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 14:19:33.503322 140206216668992 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:19:33.519575 140206216668992 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 14:19:33.519635 140206216668992 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:19:33.544291 140206216668992 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 14:19:33.544368 140206216668992 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 14:19:33.566210 140206216668992 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 14:19:33.566272 140206216668992 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 14:19:33.781609 140206216668992 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 14:19:33.781712 140206216668992 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 14:19:35.218852 140206216668992 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 14:19:35.218975 140206216668992 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 14:19:35.219031 140206216668992 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 14:19:35.219129 140206216668992 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 14:19:35.219210 140206216668992 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 14:19:35.219261 140206216668992 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 14:19:35.219354 140206216668992 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 14:19:35.219428 140206216668992 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 14:19:35.219490 140206216668992 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 14:19:35.219541 140206216668992 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 14:19:35.219609 140206216668992 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 14:19:35.219682 140206216668992 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 14:19:35.705680 140206216668992 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:19:35.705773 140206216668992 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:19:35.721066 140206216668992 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 14:19:35.721135 140206216668992 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 14:19:46.716552 140206216668992 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 14:19:46.717416 140206216668992 metric_logger.py:301] number parameters: 1.104 billion I0418 14:19:57.916868 140206216668992 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 14:19:58.091611 140206216668992 checkpointing.py:776] Waited 0.17472004890441895 seconds for step 0 to finish before starting checkpointing. I0418 14:19:58.094228 140206216668992 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:19:58.096047 140206216668992 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0 I0418 14:19:58.097433 140206216668992 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_false/checkpoints/0. I0418 14:19:58.858910 140206216668992 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 14:19:58.859961 140206216668992 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:19:58.860017 140206216668992 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 14:19:59.144579 140206216668992 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.285743s I0418 14:19:59.144767 140206216668992 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.296 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2912471294403076 s) (per-host) I0418 14:19:59.144819 140206216668992 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.291311s (batch_requests_ready=0.002314s, total_serialization_initiated=0.288926s, others=0.000071s) I0418 14:19:59.144913 140206216668992 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.295742s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.295723) I0418 14:19:59.145763 140206216668992 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.05 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_false/checkpoints/0. I0418 14:19:59.146110 140076748097280 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:39:59.146057 I0418 14:19:59.148125 140206216668992 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:19:59.148417 140074595419904 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:19:59.148571 140206216668992 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776521998.0942, 'wait_for_prev_duration_secs': 0.00011038780212402344, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776521998.0960937, 'checkpointer_blocking_duration_secs': 1.050166130065918, 'get_old_steps_start_time': 1776521999.1462882, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776521998.0921922, 'checkpoint_manager_blocking_duration_secs': 1.0563418865203857} I0418 14:19:59.148673 140206216668992 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 14:19:59.148723 140206216668992 max_utils.py:750] Memstats: After params initialized: I0418 14:19:59.148771 140206216668992 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0418 14:19:59.148804 140206216668992 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0418 14:19:59.148831 140206216668992 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0418 14:19:59.148856 140206216668992 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0418 14:19:59.459949 140206216668992 metric_logger.py:196] completed step: 0, seconds: 11.199, TFLOP/s/device: 1.213, Tokens/s/device: 182.868, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:19:59.636893 140206216668992 metric_logger.py:196] completed step: 1, seconds: 1.542, TFLOP/s/device: 8.814, Tokens/s/device: 1328.530, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 14:20:00.051684 140206216668992 metric_logger.py:196] completed step: 2, seconds: 0.029, TFLOP/s/device: 461.488, Tokens/s/device: 69560.492, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719 I0418 14:20:00.199172 140206216668992 metric_logger.py:196] completed step: 3, seconds: 0.415, TFLOP/s/device: 32.716, Tokens/s/device: 4931.256, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506 I0418 14:20:00.203626 140206216668992 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0418 14:20:00.493538 140206216668992 checkpointing.py:776] Waited 0.28988099098205566 seconds for step 5 to finish before starting checkpointing. I0418 14:20:00.496595 140206216668992 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:20:02.512523 2818 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 14:20:05.616225 140074603812608 array_metadata_store.py:203] [process=7][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_false/checkpoints/0/items/array_metadatas/process_7 I0418 14:20:09.847293 140076748097280 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 143.681 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.993744611740112 s) (per-host) I0418 14:20:09.847410 140076748097280 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.701190s. I0418 14:20:17.191647 140076748097280 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 18.045411s. I0418 14:20:17.191941 140074595419904 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:20:17.192059 140074595419904 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:20:17.192132 140074595419904 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:20:17.193715 140074595419904 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 14:20:17.193899 140206216668992 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 14:20:17.194031 140206216668992 checkpoint_manager.py:1452] Waiting for previous save to complete took 16.697438 seconds. If this number is high, consider checkpointing less frequently. I0418 14:20:17.196181 140206216668992 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 5 I0418 14:20:17.198795 140206216668992 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_false/checkpoints/5. I0418 14:20:17.919786 140206216668992 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:20:17.919909 140206216668992 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 14:20:17.961111 140206216668992 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042510s I0418 14:20:17.961292 140206216668992 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 33.536 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04599761962890625 s) (per-host) I0418 14:20:17.961352 140206216668992 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.046069s (batch_requests_ready=0.001865s, total_serialization_initiated=0.044126s, others=0.000079s) I0418 14:20:17.961467 140206216668992 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051099s (all_items=0.000015s, per_item={'items': '0.00001478'}, temp_paths=0.051084) I0418 14:20:17.962232 140206216668992 event_tracking.py:125] [process=7] [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_false/checkpoints/5. I0418 14:20:17.962541 140074595419904 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:40:17.962508 I0418 14:20:17.964583 140206216668992 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:20:17.964888 140076247930624 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:20:17.965047 140206216668992 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776522000.4965668, 'wait_for_prev_duration_secs': 16.69743800163269, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776522017.1962194, 'checkpointer_blocking_duration_secs': 0.766460657119751, 'get_old_steps_start_time': 1776522017.9627037, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776522000.494142, 'checkpoint_manager_blocking_duration_secs': 17.470869541168213} I0418 14:20:17.965180 140206216668992 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0418 14:20:17.965984 140206216668992 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.815, Tokens/s/device: 13387.108, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355 I0418 14:20:17.972104 140206216668992 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.202, Tokens/s/device: 13897.748, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467 I0418 14:20:23.966941 140074603812608 array_metadata_store.py:203] [process=7][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_false/checkpoints/5/items/array_metadatas/process_7 I0418 14:20:28.858397 140074595419904 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 144.347 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.943067789077759 s) (per-host) I0418 14:20:28.858522 140074595419904 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.895874s. I0418 14:20:35.021223 140074595419904 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 17.058558s. I0418 14:20:35.021509 140076247930624 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:20:35.021626 140076247930624 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:20:35.021675 140076247930624 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:20:35.024120 140076247930624 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0418 14:20:41.087335 140206216668992 metric_logger.py:196] completed step: 6, seconds: 17.767, TFLOP/s/device: 0.765, Tokens/s/device: 115.270, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740 I0418 14:20:41.234824 140206216668992 metric_logger.py:196] completed step: 7, seconds: 22.969, TFLOP/s/device: 0.592, Tokens/s/device: 89.163, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151 I0418 14:20:41.382266 140206216668992 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.224, Tokens/s/device: 13448.822, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098 I0418 14:20:41.529191 140206216668992 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 14:20:41.529866 140206216668992 checkpointing.py:776] Waited 0.0006973743438720703 seconds for step 9 to finish before starting checkpointing. I0418 14:20:41.532010 140206216668992 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:20:41.533645 140206216668992 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9 I0418 14:20:41.535200 140206216668992 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_125719/linen_xpk_main_20260418_125719_15_ocdbt_false/checkpoints/9. I0418 14:20:42.259689 140206216668992 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 14:20:42.259780 140206216668992 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 14:20:42.302032 140206216668992 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043207s I0418 14:20:42.302218 140206216668992 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 33.014 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.046724557876586914 s) (per-host) I0418 14:20:42.302270 140206216668992 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.046787s (batch_requests_ready=0.001754s, total_serialization_initiated=0.044966s, others=0.000067s) I0418 14:20:42.302359 140206216668992 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.050929s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.050919) I0418 14:20:42.303104 140206216668992 event_tracking.py:125] [process=7] [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_false/checkpoints/9. I0418 14:20:42.303443 140052810151680 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 14:40:42.303403 I0418 14:20:42.305386 140206216668992 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 14:20:42.305650 140055117031168 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 14:20:42.305780 140206216668992 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776522041.531978, 'wait_for_prev_duration_secs': 7.390975952148438e-05, 'time_between_consecutive_saves_sec': 6.5078184604644775, 'checkpointer_blocking_start_time': 1776522041.5336847, 'checkpointer_blocking_duration_secs': 0.7699005603790283, 'get_old_steps_start_time': 1776522042.3036027, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776522041.5301101, 'checkpoint_manager_blocking_duration_secs': 0.7756400108337402} I0418 14:20:42.305939 140206216668992 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 14:20:42.305987 140206216668992 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 14:20:49.386013 140054806632192 array_metadata_store.py:203] [process=7][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_false/checkpoints/9/items/array_metadatas/process_7 I0418 14:20:53.842370 140052810151680 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 136.327 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.586838006973267 s) (per-host) I0418 14:20:53.842492 140052810151680 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.538936s. I0418 14:21:00.690876 140052810151680 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 18.387303s. I0418 14:21:00.691180 140055117031168 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 14:21:00.691302 140055117031168 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 14:21:00.691345 140055117031168 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 14:21:00.693748 140055117031168 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 14:21:00.693927 140206216668992 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 14:21:00.694088 140206216668992 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 14:21:00.694998 140206216668992 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.151, Tokens/s/device: 13890.019, 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 14:21:11 UTC 2026 EXIT_CODE=0