XPK Start: Fri Apr 24 16:07:04 UTC 2026 2026-04-24 16:07:08.694661: 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 00:00:1777046828.707529 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777046828.711291 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777046828.722903 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046828.722921 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046828.722924 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777046828.722926 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 16:07:28.013431: 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 16:07:28.549853 136130889541440 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 16:07:37,591:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 16:07:37.591081 136130889541440 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 16:07:37,593:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-false-rs-prexx-slice-job-0-0.mt-15-ocdbt-false-rs-prexx:8482 I0424 16:07:37.593363 136130889541440 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-prexx-slice-job-0-0.mt-15-ocdbt-false-rs-prexx:8482 I0424 16:07:38.625476 136130889541440 max_utils.py:284] Jax distributed system initialized! I0424 16:07:45.622132 136130889541440 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 16:07:45.622239 136130889541440 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 16:07:45.622281 136130889541440 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 16:07:45.622315 136130889541440 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 16:07:46.242546 136130889541440 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 16:07:46.243155 136130889541440 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 16:07:46.243340 136130889541440 checkpointing.py:688] Setting up checkpoint logger... I0424 16:07:46.243391 136130889541440 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0424 16:07:46.243436 136130889541440 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 16:07:46.243800 136130889541440 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 0x7bcee3771640>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 16:07:47.567672 136130889541440 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 16:07:47.567911 136130889541440 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 0x7bba44342060>}, handler_registry=None I0424 16:07:47.568160 136130889541440 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bba44342060>` 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 16:07:47.568211 136130889541440 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bba44346cc0>` 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 16:07:47.568246 136130889541440 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 0x7bba44342060>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bba44342060>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bba44346cc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bba44346cc0>}). I0424 16:07:47.568754 136130889541440 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 16:07:47.568833 136130889541440 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7bba44138180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 16:07:49.421034 136009333339904 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777046681524992517, 'commit_timestamp_nsecs': 1777046695559825032, '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_false/checkpoints/0/_CHECKPOINT_METADATA I0424 16:07:49.680668 136009886963456 checkpoint_manager.py:1877] Missing metrics for step 9 I0424 16:07:50.475354 136130889541440 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_false/checkpoints I0424 16:07:50.477976 136130889541440 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_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bba4433f680> I0424 16:07:50.478090 136130889541440 checkpointing.py:302] Checkpoint manager created! I0424 16:07:51.373298 136130889541440 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 16:07:51.373405 136130889541440 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:07:51.743091 136130889541440 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 16:07:51.743197 136130889541440 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:07:51.759031 136130889541440 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 16:07:51.759091 136130889541440 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:07:51.781825 136130889541440 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 16:07:51.781894 136130889541440 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:07:51.797880 136130889541440 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 16:07:51.797944 136130889541440 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:07:51.813759 136130889541440 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 16:07:51.813829 136130889541440 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:07:51.838628 136130889541440 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 16:07:51.838698 136130889541440 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:07:51.858265 136130889541440 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 16:07:51.858332 136130889541440 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 16:07:52.049597 136130889541440 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 16:07:52.049710 136130889541440 checkpointing.py:582] restoring from this run's directory step 9 I0424 16:07:52.554857 136130889541440 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_false/checkpoints/9. I0424 16:07:55.528489 1461 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 16:08:03.877874 136130889541440 jax_array_handlers.py:837] [process=6] /jax/orbax/read/worker/io/requested throughput: 167.687 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 9.42397427558899 s) (per-host) I0424 16:08:03.884901 136130889541440 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/read/gbytes_per_sec: 161.287 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 9.812272310256958 s) (per-host) I0424 16:08:04.098520 136130889541440 checkpointer.py:318] Finished restoring checkpoint in 11.80 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_false/checkpoints/9. I0424 16:08:04.098693 136130889541440 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_false/checkpoints', 'checkpointer_start_time': 1777046872.2996027, 'checkpointer_duration_secs': 11.799057960510254, 'checkpoint_manager_start_time': 1777046872.0499609, 'checkpoint_manager_duration_secs': 12.048702001571655} fsdp: 32 I0424 16:08:04.506805 136130889541440 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0424 16:08:04.506938 136130889541440 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 16:08:04.506995 136130889541440 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 16:08:04.507061 136130889541440 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 16:08:04.507134 136130889541440 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 16:08:04.507179 136130889541440 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 16:08:04.507236 136130889541440 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 16:08:04.507292 136130889541440 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 16:08:04.507335 136130889541440 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 16:08:04.507375 136130889541440 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 16:08:04.507424 136130889541440 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0424 16:08:04.507471 136130889541440 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0424 16:08:05.003441 136130889541440 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 16:08:05.003537 136130889541440 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0424 16:08:05.018477 136130889541440 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 16:08:05.018542 136130889541440 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0424 16:08:15.975394 136130889541440 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 16:08:15.976188 136130889541440 metric_logger.py:289] number parameters: 1.104 billion I0424 16:08:27.153689 136130889541440 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 16:08:27.308838 136130889541440 checkpointing.py:798] Waited 0.15513110160827637 seconds for step 10 to finish before starting checkpointing. I0424 16:08:27.311297 136130889541440 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 16:08:27.313601 136130889541440 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0424 16:08:27.316204 136130889541440 async_checkpointer.py:452] [process=6] 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_false/checkpoints/10. I0424 16:08:27.875726 136130889541440 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 16:08:27.878225 136130889541440 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 16:08:27.878283 136130889541440 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 16:08:28.126987 136130889541440 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.251353s I0424 16:08:28.127170 136130889541440 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.108 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.4963967800140381 s) (per-host) I0424 16:08:28.127227 136130889541440 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.496463s (batch_requests_ready=0.242939s, total_serialization_initiated=0.253449s, others=0.000075s) I0424 16:08:28.127318 136130889541440 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.500849s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.500829) I0424 16:08:28.128215 136006632199936 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 16:08:28.128350 136130889541440 async_checkpointer.py:561] Finished blocking save. Time taken: 0.814677s. 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_false/checkpoints/10. I0424 16:08:28.567453 136130889541440 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:08:28.567829 136005038372608 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:08:28.567992 136130889541440 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046907.3112776, 'wait_for_prev_duration_secs': 6.461143493652344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777046907.3136392, 'checkpointer_blocking_duration_secs': 0.8149013519287109, 'get_old_steps_start_time': 1777046908.1285596, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1777046907.3093922, 'checkpoint_manager_blocking_duration_secs': 1.2585618495941162} I0424 16:08:28.568115 136130889541440 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 16:08:28.568170 136130889541440 max_utils.py:750] Memstats: After params initialized: I0424 16:08:28.568226 136130889541440 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 16:08:28.568265 136130889541440 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 16:08:28.568294 136130889541440 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 16:08:28.568317 136130889541440 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 16:08:28.833745 136130889541440 metric_logger.py:185] completed step: 10, seconds: 11.177, TFLOP/s/device: 1.216, Tokens/s/device: 183.227, total_weights: 65536, loss: 6.833 I0424 16:08:28.999287 136130889541440 metric_logger.py:185] completed step: 11, seconds: 1.679, TFLOP/s/device: 8.094, Tokens/s/device: 1219.970, total_weights: 65536, loss: 6.528 I0424 16:08:29.147084 136130889541440 metric_logger.py:185] completed step: 12, seconds: 0.018, TFLOP/s/device: 738.752, Tokens/s/device: 111352.762, total_weights: 65536, loss: 6.265 I0424 16:08:29.294515 136130889541440 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.940, Tokens/s/device: 13406.037, total_weights: 65536, loss: 6.045 I0424 16:08:29.298991 136130889541440 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 16:08:29.589187 136130889541440 checkpointing.py:798] Waited 0.2901623249053955 seconds for step 15 to finish before starting checkpointing. I0424 16:08:29.592185 136130889541440 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:08:32.645493 136005566850816 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_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_6 I0424 16:08:38.722790 136006632199936 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 142.409 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.09197998046875 s) (per-host) I0424 16:08:38.722912 136006632199936 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.594484s. I0424 16:08:44.273764 136006632199936 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 16.145321s. I0424 16:08:44.274029 136005038372608 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:08:44.274158 136005038372608 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:08:44.274205 136005038372608 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:08:44.275495 136005038372608 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 16:08:44.275660 136130889541440 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. W0424 16:08:44.275796 136130889541440 checkpoint_manager.py:1458] Waiting for previous save to complete took 14.683616 seconds. If this number is high, consider checkpointing less frequently. I0424 16:08:44.277956 136130889541440 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 15 I0424 16:08:44.282134 136130889541440 async_checkpointer.py:452] [process=6] 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_false/checkpoints/15. I0424 16:08:44.799107 136130889541440 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 16:08:44.799208 136130889541440 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 16:08:44.843772 136130889541440 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046161s I0424 16:08:44.843941 136130889541440 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.427 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2842586040496826 s) (per-host) I0424 16:08:44.843993 136130889541440 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284319s (batch_requests_ready=0.236164s, total_serialization_initiated=0.048086s, others=0.000068s) I0424 16:08:44.844082 136130889541440 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289152s (all_items=0.000013s, per_item={'items': '0.00001335'}, temp_paths=0.289139) I0424 16:08:44.845000 136006623807232 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 16:08:44.845184 136130889541440 async_checkpointer.py:561] Finished blocking save. Time taken: 0.567157s. 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_false/checkpoints/15. I0424 16:08:44.871448 136130889541440 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:08:44.871768 136005038372608 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:08:44.871937 136130889541440 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046909.5921564, 'wait_for_prev_duration_secs': 14.683616161346436, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777046924.277995, 'checkpointer_blocking_duration_secs': 0.5672967433929443, 'get_old_steps_start_time': 1777046924.845311, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1777046909.5898163, 'checkpoint_manager_blocking_duration_secs': 15.282084226608276} I0424 16:08:44.872041 136130889541440 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 16:08:44.872593 136130889541440 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.968, Tokens/s/device: 13862.472, total_weights: 65536, loss: 5.864 I0424 16:08:44.878270 136130889541440 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 92.069, Tokens/s/device: 13877.595, total_weights: 65536, loss: 5.719 I0424 16:09:02.864408 136130889541440 metric_logger.py:185] completed step: 16, seconds: 15.578, TFLOP/s/device: 0.872, Tokens/s/device: 131.467, total_weights: 65536, loss: 5.605 I0424 16:09:03.011955 136130889541440 metric_logger.py:185] completed step: 17, seconds: 17.840, TFLOP/s/device: 0.762, Tokens/s/device: 114.799, total_weights: 65536, loss: 5.517 I0424 16:09:03.159162 136130889541440 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.273, Tokens/s/device: 13456.244, total_weights: 65536, loss: 5.448 I0424 16:09:03.306249 136130889541440 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 16:09:03.306938 136130889541440 checkpointing.py:798] Waited 0.0007097721099853516 seconds for step 19 to finish before starting checkpointing. I0424 16:09:03.309001 136130889541440 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:09:07.443639 136005566850816 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_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_6 I0424 16:09:13.024005 136006623807232 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 55.494 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 28.46428346633911 s) (per-host) I0424 16:09:13.024140 136006623807232 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.179011s. I0424 16:09:18.450880 136006623807232 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 33.605735s. I0424 16:09:18.451208 136005038372608 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:09:18.451333 136005038372608 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:09:18.451382 136005038372608 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:09:18.452702 136005038372608 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 16:09:18.452879 136130889541440 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. W0424 16:09:18.452998 136130889541440 checkpoint_manager.py:1458] Waiting for previous save to complete took 15.144012 seconds. If this number is high, consider checkpointing less frequently. I0424 16:09:18.454349 136130889541440 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0424 16:09:18.457726 136130889541440 async_checkpointer.py:452] [process=6] 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_false/checkpoints/19. I0424 16:09:18.979168 136130889541440 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0424 16:09:18.979264 136130889541440 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 16:09:19.023122 136130889541440 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045152s I0424 16:09:19.023281 136130889541440 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.377 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2868831157684326 s) (per-host) I0424 16:09:19.023334 136130889541440 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286940s (batch_requests_ready=0.239552s, total_serialization_initiated=0.047326s, others=0.000063s) I0424 16:09:19.023422 136130889541440 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290991s (all_items=0.000011s, per_item={'items': '0.00001097'}, temp_paths=0.290980) I0424 16:09:19.024307 136005029979904 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 16:09:19.024475 136130889541440 async_checkpointer.py:561] Finished blocking save. Time taken: 0.570049s. 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_false/checkpoints/19. I0424 16:09:19.471676 136130889541440 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:09:19.472046 136005038372608 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:09:19.472210 136130889541440 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777046943.30897, 'wait_for_prev_duration_secs': 15.144011735916138, 'time_between_consecutive_saves_sec': 19.03343415260315, 'checkpointer_blocking_start_time': 1777046958.4543884, 'checkpointer_blocking_duration_secs': 0.5702192783355713, 'get_old_steps_start_time': 1777046959.024623, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1777046943.30718, 'checkpoint_manager_blocking_duration_secs': 16.16499662399292} I0424 16:09:19.472387 136130889541440 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 16:09:19.472446 136130889541440 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:09:23.850295 136005566850816 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_20260424_145744/linen_xpk_feat_nnx_set_defaults_true_20260424_145744_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_6 I0424 16:09:29.194546 136005029979904 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 151.040 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.458112478256226 s) (per-host) I0424 16:09:29.194662 136005029979904 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.170244s. I0424 16:09:34.039741 136005029979904 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 15.015309s. I0424 16:09:34.039984 136005038372608 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:09:34.040041 136005038372608 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:09:34.040086 136005038372608 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:09:34.041648 136005038372608 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 16:09:34.041801 136130889541440 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. I0424 16:09:34.041945 136130889541440 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 16:09:34.042592 136130889541440 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.131, Tokens/s/device: 13887.005, 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 16:09:44 UTC 2026 EXIT_CODE=0