XPK Start: Fri Apr 24 15:58:09 UTC 2026 2026-04-24 15:58:13.139555: 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 00:00:1777046293.152860 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777046293.156650 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777046293.168271 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046293.168291 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046293.168293 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046293.168295 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 15:58:32.958544: 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) I0424 15:58:33.488466 135367261943616 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 15:58:42,552:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 15:58:42.552900 135367261943616 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 15:58:42,555:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-true-rs-k1ikx-slice-job-0-0.mt-15-ocdbt-true-rs-k1ikx:8482 I0424 15:58:42.555247 135367261943616 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-k1ikx-slice-job-0-0.mt-15-ocdbt-true-rs-k1ikx:8482 I0424 15:58:43.023391 135367261943616 max_utils.py:284] Jax distributed system initialized! I0424 15:58:49.918213 135367261943616 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 15:58:49.918318 135367261943616 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 15:58:49.918359 135367261943616 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 I0424 15:58:49.918396 135367261943616 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 15:58:50.537869 135367261943616 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 15:58:50.538487 135367261943616 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 15:58:50.538667 135367261943616 checkpointing.py:688] Setting up checkpoint logger... I0424 15:58:50.538722 135367261943616 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 15:58:50.538766 135367261943616 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 15:58:50.539127 135367261943616 base_pytree_checkpoint_handler.py:415] 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 0x7b1d17b79130>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 15:58:51.846038 135367261943616 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 15:58:51.846278 135367261943616 checkpoint_manager.py:709] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b09ce49d730>}, handler_registry=None I0424 15:58:51.846513 135367261943616 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b09ce49d730>` 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`. I0424 15:58:51.846563 135367261943616 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b09ce4a2240>` 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`. I0424 15:58:51.846600 135367261943616 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 0x7b09ce49d730>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b09ce49d730>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b09ce4a2240>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b09ce4a2240>}). I0424 15:58:51.847122 135367261943616 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 15:58:51.847202 135367261943616 async_checkpointer.py:177] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b09cd934180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 15:58:53.362731 135245626074880 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777046168623500714, 'commit_timestamp_nsecs': 1777046213592131037, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0424 15:58:53.603331 135245626074880 checkpoint_manager.py:1877] Missing metrics for step 9 I0424 15:58:53.668191 135367261943616 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints I0424 15:58:53.998665 135367261943616 checkpoint_manager.py:929] [process=3][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_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b09ce4077a0> I0424 15:58:53.998848 135367261943616 checkpointing.py:302] Checkpoint manager created! I0424 15:58:54.900548 135367261943616 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 15:58:54.900658 135367261943616 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 15:58:55.272777 135367261943616 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 15:58:55.272891 135367261943616 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 15:58:55.292264 135367261943616 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 15:58:55.292327 135367261943616 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 15:58:55.315541 135367261943616 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 15:58:55.315611 135367261943616 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 15:58:55.334263 135367261943616 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 15:58:55.334337 135367261943616 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 15:58:55.355862 135367261943616 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 15:58:55.355950 135367261943616 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 15:58:55.382280 135367261943616 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 15:58:55.382359 135367261943616 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 15:58:55.402338 135367261943616 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 15:58:55.402404 135367261943616 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 15:58:55.595333 135367261943616 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 15:58:55.595450 135367261943616 checkpointing.py:582] restoring from this run's directory step 9 I0424 15:58:56.124785 135367261943616 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9. I0424 15:58:57.535857 1435 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 15:59:03.661066 135367261943616 jax_array_handlers.py:837] [process=3] /jax/orbax/read/worker/io/requested throughput: 257.775 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.130446195602417 s) (per-host) I0424 15:59:03.666949 135367261943616 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/read/gbytes_per_sec: 242.403 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.528784513473511 s) (per-host) I0424 15:59:04.345534 135367261943616 checkpointer.py:318] Finished restoring checkpoint in 8.49 seconds from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/9. I0424 15:59:04.345692 135367261943616 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1777046335.8584707, 'checkpointer_duration_secs': 8.48719072341919, 'checkpoint_manager_start_time': 1777046335.5957017, 'checkpoint_manager_duration_secs': 8.749961614608765} fsdp: 32 I0424 15:59:04.758323 135367261943616 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0424 15:59:04.758453 135367261943616 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) I0424 15:59:04.758508 135367261943616 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) I0424 15:59:04.758570 135367261943616 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') I0424 15:59:04.758639 135367261943616 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) I0424 15:59:04.758681 135367261943616 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) I0424 15:59:04.758743 135367261943616 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) I0424 15:59:04.758798 135367261943616 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') I0424 15:59:04.758840 135367261943616 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) I0424 15:59:04.758879 135367261943616 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) I0424 15:59:04.758941 135367261943616 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0424 15:59:04.758995 135367261943616 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0424 15:59:05.266349 135367261943616 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 15:59:05.266457 135367261943616 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0424 15:59:05.281992 135367261943616 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 15:59:05.282051 135367261943616 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0424 15:59:16.328426 135367261943616 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. I0424 15:59:16.329413 135367261943616 metric_logger.py:289] number parameters: 1.104 billion I0424 15:59:27.544665 135367261943616 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 15:59:27.703557 135367261943616 checkpointing.py:798] Waited 0.15887165069580078 seconds for step 10 to finish before starting checkpointing. I0424 15:59:27.706332 135367261943616 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 15:59:27.708322 135367261943616 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 10 I0424 15:59:27.711110 135367261943616 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10. I0424 15:59:28.304805 135367261943616 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 15:59:28.305884 135367261943616 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 15:59:28.305956 135367261943616 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 I0424 15:59:28.558693 135367261943616 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.253976s I0424 15:59:28.558866 135367261943616 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.988 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5162930488586426 s) (per-host) I0424 15:59:28.558921 135367261943616 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.516359s (batch_requests_ready=0.260282s, total_serialization_initiated=0.256003s, others=0.000074s) I0424 15:59:28.559032 135367261943616 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.520507s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.520487) I0424 15:59:28.559898 135241314322176 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 15:59:28.560086 135367261943616 async_checkpointer.py:561] Finished blocking save. Time taken: 0.851695s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10. I0424 15:59:28.849370 135367261943616 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 15:59:28.849772 135242920748800 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 15:59:28.849949 135367261943616 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046367.7063136, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777046367.7083616, 'checkpointer_blocking_duration_secs': 0.8518319129943848, 'get_old_steps_start_time': 1777046368.5602121, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1777046367.704139, 'checkpoint_manager_blocking_duration_secs': 1.145756483078003} I0424 15:59:28.850055 135367261943616 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 15:59:28.850108 135367261943616 max_utils.py:750] Memstats: After params initialized: I0424 15:59:28.850164 135367261943616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0424 15:59:28.850198 135367261943616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0424 15:59:28.850225 135367261943616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0424 15:59:28.850251 135367261943616 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0424 15:59:29.116030 135367261943616 metric_logger.py:185] completed step: 10, seconds: 11.215, TFLOP/s/device: 1.212, Tokens/s/device: 182.610, total_weights: 65536, loss: 6.833 I0424 15:59:29.287326 135367261943616 metric_logger.py:185] completed step: 11, seconds: 1.570, TFLOP/s/device: 8.654, Tokens/s/device: 1304.400, total_weights: 65536, loss: 6.528 I0424 15:59:29.435532 135367261943616 metric_logger.py:185] completed step: 12, seconds: 0.024, TFLOP/s/device: 566.036, Tokens/s/device: 85319.113, total_weights: 65536, loss: 6.265 I0424 15:59:29.582998 135367261943616 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.752, Tokens/s/device: 13377.664, total_weights: 65536, loss: 6.045 I0424 15:59:29.587547 135367261943616 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 15:59:29.877460 135367261943616 checkpointing.py:798] Waited 0.28987932205200195 seconds for step 15 to finish before starting checkpointing. I0424 15:59:29.880238 135367261943616 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 15:59:33.541223 135241305929472 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_3 I0424 16:00:09.186736 135241314322176 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 38.392 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.14412498474121 s) (per-host) I0424 16:00:09.186861 135241314322176 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.626831s. I0424 16:00:18.181081 135241314322176 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 49.621034s. I0424 16:00:18.181371 135242920748800 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:00:18.181493 135242920748800 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:00:18.181542 135242920748800 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:00:18.183235 135242920748800 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 16:00:18.183425 135367261943616 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0424 16:00:18.183576 135367261943616 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.303335 seconds. If this number is high, consider checkpointing less frequently. I0424 16:00:18.185654 135367261943616 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 15 I0424 16:00:18.189695 135367261943616 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15. I0424 16:00:19.091658 135367261943616 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 16:00:19.091759 135367261943616 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 I0424 16:00:19.135836 135367261943616 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045355s I0424 16:00:19.136025 135367261943616 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.074 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30400657653808594 s) (per-host) I0424 16:00:19.136082 135367261943616 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.304072s (batch_requests_ready=0.256757s, total_serialization_initiated=0.047242s, others=0.000072s) I0424 16:00:19.136178 135367261943616 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.309073s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.309060) I0424 16:00:19.137038 135242363762432 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 16:00:19.137166 135367261943616 async_checkpointer.py:561] Finished blocking save. Time taken: 0.951443s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15. I0424 16:00:19.150174 135367261943616 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:00:19.150417 135242920748800 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:00:19.150558 135367261943616 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046369.8802109, 'wait_for_prev_duration_secs': 48.30333495140076, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777046418.185694, 'checkpointer_blocking_duration_secs': 0.9516494274139404, 'get_old_steps_start_time': 1777046419.1373634, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1777046369.878089, 'checkpoint_manager_blocking_duration_secs': 49.272435903549194} I0424 16:00:19.150671 135367261943616 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 16:00:19.151198 135367261943616 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.671, Tokens/s/device: 13817.578, total_weights: 65536, loss: 5.864 I0424 16:00:19.156919 135367261943616 metric_logger.py:185] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.192, Tokens/s/device: 13896.145, total_weights: 65536, loss: 5.719 I0424 16:00:36.803819 135367261943616 metric_logger.py:185] completed step: 16, seconds: 49.568, TFLOP/s/device: 0.274, Tokens/s/device: 41.317, total_weights: 65536, loss: 5.605 I0424 16:00:36.951393 135367261943616 metric_logger.py:185] completed step: 17, seconds: 17.501, TFLOP/s/device: 0.776, Tokens/s/device: 117.024, total_weights: 65536, loss: 5.517 I0424 16:00:37.098629 135367261943616 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.286, Tokens/s/device: 13458.190, total_weights: 65536, loss: 5.448 I0424 16:00:37.245612 135367261943616 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 16:00:37.246310 135367261943616 checkpointing.py:798] Waited 0.0007188320159912109 seconds for step 19 to finish before starting checkpointing. I0424 16:00:37.248371 135367261943616 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:00:41.244071 135241305929472 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_3 I0424 16:01:17.653237 135242363762432 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 26.854 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 58.82117533683777 s) (per-host) I0424 16:01:17.653365 135242363762432 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 58.516123s. I0424 16:01:24.542632 135242363762432 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 65.405371s. I0424 16:01:24.542958 135242920748800 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:01:24.543086 135242920748800 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:01:24.543152 135242920748800 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:01:24.544764 135242920748800 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 16:01:24.544981 135367261943616 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0424 16:01:24.545129 135367261943616 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.296764 seconds. If this number is high, consider checkpointing less frequently. I0424 16:01:24.546561 135367261943616 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 19 I0424 16:01:24.550092 135367261943616 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19. I0424 16:01:25.472332 135367261943616 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 16:01:25.472431 135367261943616 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 I0424 16:01:25.517041 135367261943616 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045558s I0424 16:01:25.517209 135367261943616 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.989 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30916905403137207 s) (per-host) I0424 16:01:25.517260 135367261943616 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.309229s (batch_requests_ready=0.261563s, total_serialization_initiated=0.047601s, others=0.000065s) I0424 16:01:25.517351 135367261943616 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.313427s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.313416) I0424 16:01:25.518233 135245097596672 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 16:01:25.518392 135367261943616 async_checkpointer.py:561] Finished blocking save. Time taken: 0.971761s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19. I0424 16:01:25.520345 135367261943616 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:01:25.520586 135242920748800 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:01:25.520741 135367261943616 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046437.2483444, 'wait_for_prev_duration_secs': 47.29676389694214, 'time_between_consecutive_saves_sec': 19.065067529678345, 'checkpointer_blocking_start_time': 1777046484.5466, 'checkpointer_blocking_duration_secs': 0.9719061851501465, 'get_old_steps_start_time': 1777046485.5185206, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1777046437.2465475, 'checkpoint_manager_blocking_duration_secs': 48.27416133880615} I0424 16:01:25.520874 135367261943616 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 16:01:25.520920 135367261943616 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:01:30.311149 135245584111360 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_3 I0424 16:02:06.898443 135245097596672 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 37.889 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.690362215042114 s) (per-host) I0424 16:02:06.898564 135245097596672 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.380220s. I0424 16:02:13.207852 135245097596672 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 47.689492s. I0424 16:02:13.208149 135242920748800 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:02:13.208274 135242920748800 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:02:13.208320 135242920748800 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:02:13.209898 135242920748800 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 16:02:13.210097 135367261943616 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0424 16:02:13.210240 135367261943616 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 16:02:13.210896 135367261943616 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.123, Tokens/s/device: 13885.781, 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: Fri Apr 24 16:02:24 UTC 2026 EXIT_CODE=0