XPK Start: Wed Apr 22 16:58:39 UTC 2026 2026-04-22 16:58:43.652120: 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 00:00:1776877123.665266 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776877123.668999 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776877123.680564 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776877123.680587 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776877123.680589 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776877123.680591 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-22 16:59:02.789150: 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) I0422 16:59:03.312624 133064994481984 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-22 16:59:12,353:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0422 16:59:12.353299 133064994481984 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-22 16:59:12,355:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-true-rs-ol77s-slice-job-0-0.mt-15-ocdbt-true-rs-ol77s:8482 I0422 16:59:12.355700 133064994481984 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-ol77s-slice-job-0-0.mt-15-ocdbt-true-rs-ol77s:8482 I0422 16:59:13.438445 133064994481984 max_utils.py:284] Jax distributed system initialized! I0422 16:59:20.442173 133064994481984 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 16:59:20.442278 133064994481984 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 16:59:20.442319 133064994481984 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 I0422 16:59:20.442354 133064994481984 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 16:59:21.067365 133064994481984 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 16:59:21.067980 133064994481984 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 16:59:21.068164 133064994481984 checkpointing.py:688] Setting up checkpoint logger... I0422 16:59:21.068215 133064994481984 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 16:59:21.068258 133064994481984 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0422 16:59:21.068608 133064994481984 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 0x790519d71490>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 16:59:22.326690 133064994481984 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 16:59:22.326936 133064994481984 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 0x78f0d8211e80>}, handler_registry=None I0422 16:59:22.327206 133064994481984 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f0d8211e80>` 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`. I0422 16:59:22.327262 133064994481984 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f0d8216780>` 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`. I0422 16:59:22.327302 133064994481984 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 0x78f0d8211e80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f0d8211e80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f0d8216780>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f0d8216780>}). I0422 16:59:22.327921 133064994481984 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0422 16:59:22.328033 133064994481984 async_checkpointer.py:177] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78f0d81b8180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 16:59:23.990056 132945159362304 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776876879347229571, 'commit_timestamp_nsecs': 1776876921689801392, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0422 16:59:24.241106 132945159362304 checkpoint_manager.py:1877] Missing metrics for step 0 I0422 16:59:24.474954 133064994481984 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints I0422 16:59:24.880328 133064994481984 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_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78f0d8212fc0> I0422 16:59:24.880520 133064994481984 checkpointing.py:302] Checkpoint manager created! I0422 16:59:25.790419 133064994481984 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 16:59:25.790526 133064994481984 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 16:59:26.160470 133064994481984 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 16:59:26.160564 133064994481984 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 16:59:26.176772 133064994481984 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 16:59:26.176835 133064994481984 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 16:59:26.199907 133064994481984 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 16:59:26.199991 133064994481984 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 16:59:26.216558 133064994481984 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 16:59:26.216620 133064994481984 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 16:59:26.232844 133064994481984 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 16:59:26.232907 133064994481984 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 16:59:26.257939 133064994481984 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 16:59:26.258013 133064994481984 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 16:59:26.277799 133064994481984 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 16:59:26.277863 133064994481984 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 16:59:26.467987 133064994481984 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 16:59:26.468098 133064994481984 checkpointing.py:582] restoring from this run's directory step 9 I0422 16:59:26.971056 133064994481984 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/9. I0422 16:59:29.256743 1414 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 16:59:35.453437 133064994481984 jax_array_handlers.py:837] [process=3] /jax/orbax/read/worker/io/requested throughput: 254.801 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.20198130607605 s) (per-host) I0422 16:59:35.469721 133064994481984 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/read/gbytes_per_sec: 224.597 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.046362400054932 s) (per-host) I0422 16:59:35.588629 133064994481984 checkpointer.py:318] Finished restoring checkpoint in 8.87 seconds from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/9. I0422 16:59:35.588795 133064994481984 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776877166.721003, 'checkpointer_duration_secs': 8.867760181427002, 'checkpoint_manager_start_time': 1776877166.468376, 'checkpoint_manager_duration_secs': 9.120389461517334} fsdp: 32 I0422 16:59:35.999245 133064994481984 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 16:59:35.999379 133064994481984 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) I0422 16:59:35.999434 133064994481984 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) I0422 16:59:35.999499 133064994481984 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') I0422 16:59:35.999554 133064994481984 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) I0422 16:59:35.999596 133064994481984 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) I0422 16:59:35.999650 133064994481984 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) I0422 16:59:35.999703 133064994481984 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') I0422 16:59:35.999745 133064994481984 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) I0422 16:59:35.999784 133064994481984 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) I0422 16:59:35.999833 133064994481984 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0422 16:59:35.999881 133064994481984 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0422 16:59:36.498058 133064994481984 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 16:59:36.498153 133064994481984 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0422 16:59:36.513115 133064994481984 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 16:59:36.513174 133064994481984 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0422 16:59:47.363358 133064994481984 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. I0422 16:59:47.364304 133064994481984 metric_logger.py:289] number parameters: 1.104 billion I0422 16:59:58.631284 133064994481984 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0422 16:59:58.785965 133064994481984 checkpointing.py:798] Waited 0.15463995933532715 seconds for step 10 to finish before starting checkpointing. I0422 16:59:58.788375 133064994481984 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 16:59:58.790328 133064994481984 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 10 I0422 16:59:58.793613 133064994481984 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/10. I0422 16:59:59.850740 133064994481984 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 16:59:59.851892 133064994481984 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 16:59:59.851962 133064994481984 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 I0422 17:00:00.102298 133064994481984 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.251650s I0422 17:00:00.102467 133064994481984 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.742 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5625600814819336 s) (per-host) I0422 17:00:00.102519 133064994481984 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.562622s (batch_requests_ready=0.308872s, total_serialization_initiated=0.253679s, others=0.000070s) I0422 17:00:00.102608 133064994481984 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.566792s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.566772) I0422 17:00:00.103532 132944572167936 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:00:00.103701 133064994481984 async_checkpointer.py:561] Finished blocking save. Time taken: 1.313295s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/10. I0422 17:00:00.289300 133064994481984 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:00:00.289662 132940277200640 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:00:00.289829 133064994481984 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776877198.7883563, 'wait_for_prev_duration_secs': 6.747245788574219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776877198.790368, 'checkpointer_blocking_duration_secs': 1.313443660736084, 'get_old_steps_start_time': 1776877200.1038318, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776877198.7865167, 'checkpoint_manager_blocking_duration_secs': 1.5032737255096436} I0422 17:00:00.289954 133064994481984 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0422 17:00:00.290010 133064994481984 max_utils.py:750] Memstats: After params initialized: I0422 17:00:00.290069 133064994481984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0422 17:00:00.290106 133064994481984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0422 17:00:00.290135 133064994481984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0422 17:00:00.290161 133064994481984 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0422 17:00:00.555643 133064994481984 metric_logger.py:185] completed step: 10, seconds: 11.267, TFLOP/s/device: 1.206, Tokens/s/device: 181.772, total_weights: 65536, loss: 6.833 I0422 17:00:00.723978 133064994481984 metric_logger.py:185] completed step: 11, seconds: 1.923, TFLOP/s/device: 7.066, Tokens/s/device: 1065.054, total_weights: 65536, loss: 6.528 I0422 17:00:00.871881 133064994481984 metric_logger.py:185] completed step: 12, seconds: 0.021, TFLOP/s/device: 636.668, Tokens/s/device: 95965.512, total_weights: 65536, loss: 6.265 I0422 17:00:01.019135 133064994481984 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.830, Tokens/s/device: 13389.471, total_weights: 65536, loss: 6.045 I0422 17:00:01.023621 133064994481984 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0422 17:00:01.313820 133064994481984 checkpointing.py:798] Waited 0.2901742458343506 seconds for step 15 to finish before starting checkpointing. I0422 17:00:01.316568 133064994481984 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:00:05.163838 132940293986048 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_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_3 I0422 17:00:40.321152 132944572167936 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 38.733 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.78120398521423 s) (per-host) I0422 17:00:40.321276 132944572167936 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.217633s. I0422 17:00:48.380247 132944572167936 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 48.276590s. I0422 17:00:48.380480 132940277200640 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:00:48.380548 132940277200640 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:00:48.380592 132940277200640 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:00:48.382206 132940277200640 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 17:00:48.382323 133064994481984 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. W0422 17:00:48.382396 133064994481984 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.065844 seconds. If this number is high, consider checkpointing less frequently. I0422 17:00:48.384711 133064994481984 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 15 I0422 17:00:48.388896 133064994481984 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/15. I0422 17:00:48.955043 133064994481984 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 17:00:48.955145 133064994481984 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 I0422 17:00:48.999096 133064994481984 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045196s I0422 17:00:48.999270 133064994481984 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.073 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3040645122528076 s) (per-host) I0422 17:00:48.999322 133064994481984 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.304126s (batch_requests_ready=0.256981s, total_serialization_initiated=0.047075s, others=0.000070s) I0422 17:00:48.999414 133064994481984 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.309000s (all_items=0.000013s, per_item={'items': '0.00001287'}, temp_paths=0.308987) I0422 17:00:49.000282 132940293986048 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:00:49.000443 133064994481984 async_checkpointer.py:561] Finished blocking save. Time taken: 0.615653s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/15. I0422 17:00:49.030433 133064994481984 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:00:49.030712 132940277200640 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:00:49.030853 133064994481984 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776877201.316532, 'wait_for_prev_duration_secs': 47.06584358215332, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776877248.3847585, 'checkpointer_blocking_duration_secs': 0.6158037185668945, 'get_old_steps_start_time': 1776877249.0005841, 'get_old_steps_duration_secs': 2.3603439331054688e-05, 'checkpoint_manager_blocking_start_time': 1776877201.3144248, 'checkpoint_manager_blocking_duration_secs': 47.71639323234558} I0422 17:00:49.030978 133064994481984 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0422 17:00:49.031496 133064994481984 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.851, Tokens/s/device: 13844.760, total_weights: 65536, loss: 5.864 I0422 17:00:49.037674 133064994481984 metric_logger.py:185] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.367, Tokens/s/device: 13922.596, total_weights: 65536, loss: 5.719 I0422 17:01:06.957957 133064994481984 metric_logger.py:185] completed step: 16, seconds: 48.013, TFLOP/s/device: 0.283, Tokens/s/device: 42.655, total_weights: 65536, loss: 5.605 I0422 17:01:07.105217 133064994481984 metric_logger.py:185] completed step: 17, seconds: 17.774, TFLOP/s/device: 0.764, Tokens/s/device: 115.223, total_weights: 65536, loss: 5.517 I0422 17:01:07.252627 133064994481984 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.099, Tokens/s/device: 13429.949, total_weights: 65536, loss: 5.448 I0422 17:01:07.399625 133064994481984 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0422 17:01:07.400308 133064994481984 checkpointing.py:798] Waited 0.000701904296875 seconds for step 19 to finish before starting checkpointing. I0422 17:01:07.402266 133064994481984 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:01:11.455662 132941871027968 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_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_3 I0422 17:01:48.455018 132940293986048 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 26.432 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 59.75977182388306 s) (per-host) I0422 17:01:48.455143 132940293986048 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 59.454745s. I0422 17:01:56.888350 132940293986048 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 67.887939s. I0422 17:01:56.888594 132940277200640 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:01:56.888660 132940277200640 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:01:56.888716 132940277200640 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:01:56.891480 132940277200640 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 17:01:56.891652 133064994481984 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. W0422 17:01:56.891793 133064994481984 checkpoint_manager.py:1458] Waiting for previous save to complete took 49.489540 seconds. If this number is high, consider checkpointing less frequently. I0422 17:01:56.893978 133064994481984 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 19 I0422 17:01:56.898216 133064994481984 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/19. I0422 17:01:57.601386 133064994481984 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0422 17:01:57.601487 133064994481984 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 I0422 17:01:57.647567 133064994481984 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.047221s I0422 17:01:57.647735 133064994481984 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.043 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30587196350097656 s) (per-host) I0422 17:01:57.647785 133064994481984 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.305931s (batch_requests_ready=0.256592s, total_serialization_initiated=0.049275s, others=0.000064s) I0422 17:01:57.647874 133064994481984 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.310012s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.310002) I0422 17:01:57.648847 132944563775232 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0422 17:01:57.649036 133064994481984 async_checkpointer.py:561] Finished blocking save. Time taken: 0.754987s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/19. I0422 17:01:57.660371 133064994481984 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:01:57.660602 132940277200640 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:01:57.660739 133064994481984 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776877267.402237, 'wait_for_prev_duration_secs': 49.489540100097656, 'time_between_consecutive_saves_sec': 19.019986867904663, 'checkpointer_blocking_start_time': 1776877316.8940172, 'checkpointer_blocking_duration_secs': 0.7551307678222656, 'get_old_steps_start_time': 1776877317.6491606, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1776877267.4005435, 'checkpoint_manager_blocking_duration_secs': 50.26015877723694} I0422 17:01:57.660842 133064994481984 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0422 17:01:57.660885 133064994481984 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:02:02.437133 132944572167936 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_20260422_154647/linen_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_3 I0422 17:02:38.408842 132944563775232 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 38.464 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.06693625450134 s) (per-host) I0422 17:02:38.408965 132944563775232 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.759991s. I0422 17:02:47.083445 132944563775232 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 49.434456s. I0422 17:02:47.083729 132940277200640 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:02:47.083875 132940277200640 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:02:47.083916 132940277200640 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:02:47.085361 132940277200640 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 17:02:47.085543 133064994481984 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. I0422 17:02:47.085689 133064994481984 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:02:47.086382 133064994481984 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.262, Tokens/s/device: 13906.808, 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: Wed Apr 22 17:02:59 UTC 2026 EXIT_CODE=0