XPK Start: Thu Apr 23 17:09:10 UTC 2026 2026-04-23 17:09:14.929058: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776964154.942301 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776964154.946046 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776964154.957698 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776964154.957716 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776964154.957718 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776964154.957720 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 17:09:34.232204: 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) I0423 17:09:34.758938 138277507901248 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 17:09:43,802:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 17:09:43.802563 138277507901248 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 17:09:43,804:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-false-rs-lf3j0-slice-job-0-0.mt-15-ocdbt-false-rs-lf3j0:8482 I0423 17:09:43.804952 138277507901248 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-lf3j0-slice-job-0-0.mt-15-ocdbt-false-rs-lf3j0:8482 I0423 17:09:44.712236 138277507901248 max_utils.py:284] Jax distributed system initialized! I0423 17:09:51.455685 138277507901248 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 17:09:51.455795 138277507901248 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 17:09:51.455836 138277507901248 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0423 17:09:51.455870 138277507901248 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 17:09:52.092615 138277507901248 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 17:09:52.093299 138277507901248 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 17:09:52.093515 138277507901248 checkpointing.py:688] Setting up checkpoint logger... I0423 17:09:52.093572 138277507901248 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0423 17:09:52.093633 138277507901248 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 17:09:52.094013 138277507901248 base_pytree_checkpoint_handler.py:415] 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 0x7dc2b1d71130>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 17:09:53.433028 138277507901248 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 17:09:53.433423 138277507901248 checkpoint_manager.py:709] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7daf56d9dca0>}, handler_registry=None I0423 17:09:53.433663 138277507901248 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7daf56d9dca0>` 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`. I0423 17:09:53.433716 138277507901248 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7daf56d089e0>` 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`. I0423 17:09:53.433755 138277507901248 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 0x7daf56d9dca0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7daf56d9dca0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7daf56d089e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7daf56d089e0>}). I0423 17:09:53.434271 138277507901248 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 17:09:53.434350 138277507901248 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7daf55460180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 17:09:55.104236 138156431087360 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776964010588205231, 'commit_timestamp_nsecs': 1776964024256823221, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0423 17:09:55.356436 138155869071104 checkpoint_manager.py:1877] Missing metrics for step 9 I0423 17:09:55.370118 138277507901248 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints I0423 17:09:55.700496 138277507901248 checkpoint_manager.py:929] [process=6][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_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7daf570a7d70> I0423 17:09:55.700677 138277507901248 checkpointing.py:302] Checkpoint manager created! I0423 17:09:56.609737 138277507901248 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:09:56.609844 138277507901248 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:09:56.980610 138277507901248 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:09:56.980703 138277507901248 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:09:56.996697 138277507901248 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:09:56.996758 138277507901248 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:09:57.019541 138277507901248 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:09:57.019608 138277507901248 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:09:57.035551 138277507901248 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:09:57.035612 138277507901248 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:09:57.051351 138277507901248 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:09:57.051412 138277507901248 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:09:57.076406 138277507901248 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 17:09:57.076489 138277507901248 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:09:57.095934 138277507901248 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 17:09:57.095999 138277507901248 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 17:09:57.286000 138277507901248 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 17:09:57.286131 138277507901248 checkpointing.py:582] restoring from this run's directory step 9 I0423 17:09:57.770078 138277507901248 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/9. I0423 17:10:00.254072 1465 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 17:10:09.660335 138277507901248 jax_array_handlers.py:837] [process=6] /jax/orbax/read/worker/io/requested throughput: 152.899 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.335406303405762 s) (per-host) I0423 17:10:09.677782 138277507901248 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/read/gbytes_per_sec: 146.878 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.774884462356567 s) (per-host) I0423 17:10:10.085307 138277507901248 checkpointer.py:318] Finished restoring checkpoint in 12.56 seconds from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/9. I0423 17:10:10.085479 138277507901248 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776964197.5268893, 'checkpointer_duration_secs': 12.558558702468872, 'checkpoint_manager_start_time': 1776964197.2863932, 'checkpoint_manager_duration_secs': 12.799057006835938} fsdp: 32 I0423 17:10:10.490926 138277507901248 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 17:10:10.491080 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 17:10:10.491151 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 17:10:10.491234 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0423 17:10:10.491316 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 17:10:10.491396 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 17:10:10.491491 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 17:10:10.491568 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0423 17:10:10.491635 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0423 17:10:10.491694 138277507901248 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 17:10:10.491778 138277507901248 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0423 17:10:10.491863 138277507901248 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0423 17:10:10.989667 138277507901248 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0423 17:10:10.989762 138277507901248 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0423 17:10:11.005174 138277507901248 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0423 17:10:11.005234 138277507901248 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0423 17:10:21.828953 138277507901248 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. I0423 17:10:21.829893 138277507901248 metric_logger.py:289] number parameters: 1.104 billion I0423 17:10:33.055187 138277507901248 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0423 17:10:33.208624 138277507901248 checkpointing.py:798] Waited 0.15341901779174805 seconds for step 10 to finish before starting checkpointing. I0423 17:10:33.211042 138277507901248 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:10:33.212964 138277507901248 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0423 17:10:33.215951 138277507901248 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/10. I0423 17:10:33.799683 138277507901248 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 17:10:33.800736 138277507901248 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0423 17:10:33.800800 138277507901248 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 I0423 17:10:34.048991 138277507901248 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.249397s I0423 17:10:34.049191 138277507901248 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.057 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5046515464782715 s) (per-host) I0423 17:10:34.049257 138277507901248 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.504729s (batch_requests_ready=0.253213s, total_serialization_initiated=0.251430s, others=0.000086s) I0423 17:10:34.049354 138277507901248 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.509444s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.509424) I0423 17:10:34.050270 138151557318400 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:10:34.050440 138277507901248 async_checkpointer.py:561] Finished blocking save. Time taken: 0.837404s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/10. I0423 17:10:34.098920 138277507901248 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:10:34.099302 138153138566912 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:10:34.099475 138277507901248 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776964233.2110236, 'wait_for_prev_duration_secs': 6.341934204101562e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776964233.213003, 'checkpointer_blocking_duration_secs': 0.8375487327575684, 'get_old_steps_start_time': 1776964234.0505695, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776964233.209224, 'checkpoint_manager_blocking_duration_secs': 0.890216588973999} I0423 17:10:34.099587 138277507901248 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0423 17:10:34.099639 138277507901248 max_utils.py:750] Memstats: After params initialized: I0423 17:10:34.099706 138277507901248 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 17:10:34.099739 138277507901248 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 17:10:34.099766 138277507901248 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 17:10:34.099790 138277507901248 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 17:10:34.366478 138277507901248 metric_logger.py:185] completed step: 10, seconds: 11.225, TFLOP/s/device: 1.210, Tokens/s/device: 182.447, total_weights: 65536, loss: 6.833 I0423 17:10:34.538409 138277507901248 metric_logger.py:185] completed step: 11, seconds: 1.310, TFLOP/s/device: 10.372, Tokens/s/device: 1563.420, total_weights: 65536, loss: 6.528 I0423 17:10:34.686002 138277507901248 metric_logger.py:185] completed step: 12, seconds: 0.025, TFLOP/s/device: 551.426, Tokens/s/device: 83116.883, total_weights: 65536, loss: 6.265 I0423 17:10:34.833346 138277507901248 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.677, Tokens/s/device: 13366.314, total_weights: 65536, loss: 6.045 I0423 17:10:34.838123 138277507901248 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0423 17:10:35.127989 138277507901248 checkpointing.py:798] Waited 0.28983330726623535 seconds for step 15 to finish before starting checkpointing. I0423 17:10:35.130828 138277507901248 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:10:38.677711 138150996735744 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_6 I0423 17:10:44.685009 138151557318400 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 141.789 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.14044713973999 s) (per-host) I0423 17:10:44.685132 138151557318400 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.634748s. I0423 17:10:50.009475 138151557318400 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 15.959069s. I0423 17:10:50.009771 138153138566912 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:10:50.009886 138153138566912 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:10:50.009936 138153138566912 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:10:50.012460 138153138566912 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 17:10:50.012639 138277507901248 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0423 17:10:50.012768 138277507901248 checkpoint_manager.py:1458] Waiting for previous save to complete took 14.881945 seconds. If this number is high, consider checkpointing less frequently. I0423 17:10:50.014706 138277507901248 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 15 I0423 17:10:50.018725 138277507901248 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/15. I0423 17:10:50.546198 138277507901248 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0423 17:10:50.546299 138277507901248 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 I0423 17:10:50.588773 138277507901248 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044033s I0423 17:10:50.588945 138277507901248 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.313 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2903561592102051 s) (per-host) I0423 17:10:50.589000 138277507901248 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290421s (batch_requests_ready=0.244437s, total_serialization_initiated=0.045910s, others=0.000074s) I0423 17:10:50.589089 138277507901248 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.295324s (all_items=0.000013s, per_item={'items': '0.00001335'}, temp_paths=0.295310) I0423 17:10:50.589962 138148889749248 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:10:50.590138 138277507901248 async_checkpointer.py:561] Finished blocking save. Time taken: 0.575357s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/15. I0423 17:10:50.640073 138277507901248 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:10:50.640406 138153138566912 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:10:50.640567 138277507901248 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776964235.1307993, 'wait_for_prev_duration_secs': 14.881945133209229, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776964250.0147462, 'checkpointer_blocking_duration_secs': 0.5754997730255127, 'get_old_steps_start_time': 1776964250.590265, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776964235.1286945, 'checkpoint_manager_blocking_duration_secs': 15.511839389801025} I0423 17:10:50.640683 138277507901248 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0423 17:10:50.641292 138277507901248 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.107, Tokens/s/device: 13883.334, total_weights: 65536, loss: 5.864 I0423 17:10:50.647027 138277507901248 metric_logger.py:185] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.132, Tokens/s/device: 13887.194, total_weights: 65536, loss: 5.719 I0423 17:11:08.226359 138277507901248 metric_logger.py:185] completed step: 16, seconds: 15.808, TFLOP/s/device: 0.860, Tokens/s/device: 129.557, total_weights: 65536, loss: 5.605 I0423 17:11:08.373985 138277507901248 metric_logger.py:185] completed step: 17, seconds: 17.433, TFLOP/s/device: 0.779, Tokens/s/device: 117.478, total_weights: 65536, loss: 5.517 I0423 17:11:08.521363 138277507901248 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.287, Tokens/s/device: 13458.278, total_weights: 65536, loss: 5.448 I0423 17:11:08.668477 138277507901248 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0423 17:11:08.669162 138277507901248 checkpointing.py:798] Waited 0.0007033348083496094 seconds for step 19 to finish before starting checkpointing. I0423 17:11:08.671002 138277507901248 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:11:12.902917 138150996735744 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_6 I0423 17:11:19.484683 138148889749248 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 54.122 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 29.186055660247803 s) (per-host) I0423 17:11:19.484806 138148889749248 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.894732s. I0423 17:11:23.352566 138148889749248 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 32.762478s. I0423 17:11:23.352822 138153138566912 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:11:23.352935 138153138566912 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:11:23.352982 138153138566912 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:11:23.354617 138153138566912 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0423 17:11:23.354790 138277507901248 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0423 17:11:23.354905 138277507901248 checkpoint_manager.py:1458] Waiting for previous save to complete took 14.683916 seconds. If this number is high, consider checkpointing less frequently. I0423 17:11:23.356455 138277507901248 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0423 17:11:23.359799 138277507901248 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/19. I0423 17:11:24.177649 138277507901248 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0423 17:11:24.177743 138277507901248 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 I0423 17:11:24.221496 138277507901248 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044855s I0423 17:11:24.221669 138277507901248 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.230 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2949235439300537 s) (per-host) I0423 17:11:24.221718 138277507901248 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.294983s (batch_requests_ready=0.247941s, total_serialization_initiated=0.046978s, others=0.000064s) I0423 17:11:24.221807 138277507901248 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.298952s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.298940) I0423 17:11:24.222629 138149393049344 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:11:24.222781 138277507901248 async_checkpointer.py:561] Finished blocking save. Time taken: 0.866257s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/19. I0423 17:11:24.654925 138277507901248 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:11:24.655312 138153138566912 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:11:24.655457 138277507901248 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776964268.670974, 'wait_for_prev_duration_secs': 14.683916330337524, 'time_between_consecutive_saves_sec': 18.658472061157227, 'checkpointer_blocking_start_time': 1776964283.356495, 'checkpointer_blocking_duration_secs': 0.8663949966430664, 'get_old_steps_start_time': 1776964284.2229028, 'get_old_steps_duration_secs': 2.0265579223632812e-05, 'checkpoint_manager_blocking_start_time': 1776964268.6693876, 'checkpoint_manager_blocking_duration_secs': 15.986035585403442} I0423 17:11:24.655626 138277507901248 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0423 17:11:24.655667 138277507901248 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:11:28.984713 138150996735744 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/linen_xpk_feat_nnx_set_defaults_true_20260423_155242_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_6 I0423 17:11:35.069624 138149393049344 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 141.759 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.142837285995483 s) (per-host) I0423 17:11:35.069737 138149393049344 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.847001s. I0423 17:11:39.163043 138149393049344 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 14.940292s. I0423 17:11:39.163385 138153138566912 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:11:39.163502 138153138566912 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:11:39.163542 138153138566912 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:11:39.164988 138153138566912 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 17:11:39.165184 138277507901248 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0423 17:11:39.165356 138277507901248 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:11:39.166072 138277507901248 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.134, Tokens/s/device: 13887.476, total_weights: 65536, loss: 5.391 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 23 17:11:51 UTC 2026 EXIT_CODE=0