XPK Start: Mon Apr 20 20:39:39 UTC 2026 2026-04-20 20:39:43.957042: 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 00:00:1776717583.969710 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776717583.973372 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776717583.984550 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776717583.984569 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776717583.984572 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776717583.984573 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 20:40:03.308049: 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) I0420 20:40:03.837577 135309809014592 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 20:40:12,879:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 20:40:12.879296 135309809014592 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 20:40:12,881:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-true-rs-efbxt-slice-job-0-0.mt-16-zarr3-true-rs-efbxt:8482 I0420 20:40:12.881680 135309809014592 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-true-rs-efbxt-slice-job-0-0.mt-16-zarr3-true-rs-efbxt:8482 I0420 20:40:13.692985 135309809014592 max_utils.py:284] Jax distributed system initialized! I0420 20:40:19.671410 135309809014592 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 20:40:19.671516 135309809014592 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 20:40:19.671557 135309809014592 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 I0420 20:40:19.671592 135309809014592 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 20:40:20.293298 135309809014592 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 20:40:20.293900 135309809014592 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 20:40:20.294094 135309809014592 checkpointing.py:688] Setting up checkpoint logger... I0420 20:40:20.294159 135309809014592 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 20:40:20.294205 135309809014592 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 20:40:20.294552 135309809014592 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 0x7b0fc0b54fb0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 20:40:21.539500 135309809014592 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 20:40:21.539759 135309809014592 checkpoint_manager.py:709] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afd4f5e5880>}, handler_registry=None I0420 20:40:21.540005 135309809014592 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afd4f5e5880>` 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`. I0420 20:40:21.540055 135309809014592 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afd4f5ea7b0>` 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`. I0420 20:40:21.540091 135309809014592 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 0x7afd4f5e5880>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afd4f5e5880>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afd4f5ea7b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afd4f5ea7b0>}). I0420 20:40:21.540615 135309809014592 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 20:40:21.540713 135309809014592 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7afd4eec4180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 20:40:23.506839 135192341636864 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776717459732273339, 'commit_timestamp_nsecs': 1776717506438006035, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0420 20:40:23.759880 135192341636864 checkpoint_manager.py:1877] Missing metrics for step 9 I0420 20:40:24.190182 135309809014592 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints I0420 20:40:24.261007 135309809014592 checkpoint_manager.py:929] [process=7][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7afdc4410860> I0420 20:40:24.261180 135309809014592 checkpointing.py:302] Checkpoint manager created! I0420 20:40:25.161621 135309809014592 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 20:40:25.161748 135309809014592 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 20:40:25.534308 135309809014592 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 20:40:25.534400 135309809014592 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 20:40:25.550539 135309809014592 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 20:40:25.550600 135309809014592 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 20:40:25.573474 135309809014592 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 20:40:25.573541 135309809014592 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 20:40:25.589400 135309809014592 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 20:40:25.589463 135309809014592 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 20:40:25.605233 135309809014592 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 20:40:25.605294 135309809014592 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 20:40:25.630161 135309809014592 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 20:40:25.630234 135309809014592 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 20:40:25.649868 135309809014592 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 20:40:25.649934 135309809014592 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 20:40:25.839145 135309809014592 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 20:40:25.839260 135309809014592 checkpointing.py:582] restoring from this run's directory step 9 I0420 20:40:26.339534 135309809014592 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/9. I0420 20:40:27.777525 1374 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 20:40:34.026144 135309809014592 jax_array_handlers.py:837] [process=7] /jax/orbax/read/worker/io/requested throughput: 252.689 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.253826379776001 s) (per-host) I0420 20:40:34.040462 135309809014592 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/read/gbytes_per_sec: 236.440 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.6934287548065186 s) (per-host) I0420 20:40:34.837734 135309809014592 checkpointer.py:318] Finished restoring checkpoint in 8.75 seconds from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/9. I0420 20:40:34.837892 135309809014592 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776717626.0884347, 'checkpointer_duration_secs': 8.749425172805786, 'checkpoint_manager_start_time': 1776717625.8395042, 'checkpoint_manager_duration_secs': 8.99835753440857} fsdp: 32 I0420 20:40:35.240988 135309809014592 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0420 20:40:35.241119 135309809014592 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) I0420 20:40:35.241173 135309809014592 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) I0420 20:40:35.241235 135309809014592 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') I0420 20:40:35.241295 135309809014592 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) I0420 20:40:35.241339 135309809014592 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) I0420 20:40:35.241392 135309809014592 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) I0420 20:40:35.241448 135309809014592 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') I0420 20:40:35.241490 135309809014592 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) I0420 20:40:35.241529 135309809014592 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) I0420 20:40:35.241576 135309809014592 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0420 20:40:35.241623 135309809014592 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0420 20:40:35.738780 135309809014592 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 20:40:35.738875 135309809014592 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 20:40:35.753786 135309809014592 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 20:40:35.753845 135309809014592 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 20:40:46.543813 135309809014592 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. I0420 20:40:46.544830 135309809014592 metric_logger.py:289] number parameters: 1.104 billion I0420 20:40:57.666921 135309809014592 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 20:40:57.863736 135309809014592 checkpointing.py:798] Waited 0.19680094718933105 seconds for step 10 to finish before starting checkpointing. I0420 20:40:57.866248 135309809014592 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 20:40:57.868230 135309809014592 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 10 I0420 20:40:57.871109 135309809014592 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/10. I0420 20:40:58.911954 135309809014592 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 20:40:58.914502 135309809014592 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0420 20:40:58.914564 135309809014592 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 I0420 20:40:59.167385 135309809014592 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.255521s I0420 20:40:59.167545 135309809014592 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.850 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5412607192993164 s) (per-host) I0420 20:40:59.167594 135309809014592 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.541318s (batch_requests_ready=0.283714s, total_serialization_initiated=0.257539s, others=0.000066s) I0420 20:40:59.167689 135309809014592 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.545441s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.545422) I0420 20:40:59.168611 135192316458752 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 20:40:59.168795 135309809014592 async_checkpointer.py:561] Finished blocking save. Time taken: 1.300493s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/10. I0420 20:40:59.170742 135309809014592 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 20:40:59.171060 135186830317312 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 20:40:59.171215 135309809014592 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776717657.8662293, 'wait_for_prev_duration_secs': 6.556510925292969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776717657.8682687, 'checkpointer_blocking_duration_secs': 1.3006343841552734, 'get_old_steps_start_time': 1776717659.1689208, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1776717657.864324, 'checkpoint_manager_blocking_duration_secs': 1.3068552017211914} I0420 20:40:59.171323 135309809014592 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 20:40:59.171374 135309809014592 max_utils.py:750] Memstats: After params initialized: I0420 20:40:59.171422 135309809014592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0420 20:40:59.171455 135309809014592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0420 20:40:59.171483 135309809014592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0420 20:40:59.171508 135309809014592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0420 20:40:59.438649 135309809014592 metric_logger.py:185] completed step: 10, seconds: 11.122, TFLOP/s/device: 1.222, Tokens/s/device: 184.140, total_weights: 65536, loss: 6.833 I0420 20:40:59.604464 135309809014592 metric_logger.py:185] completed step: 11, seconds: 1.770, TFLOP/s/device: 7.675, Tokens/s/device: 1156.828, total_weights: 65536, loss: 6.528 I0420 20:40:59.752122 135309809014592 metric_logger.py:185] completed step: 12, seconds: 0.019, TFLOP/s/device: 726.429, Tokens/s/device: 109495.295, total_weights: 65536, loss: 6.265 I0420 20:40:59.899693 135309809014592 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.811, Tokens/s/device: 13386.583, total_weights: 65536, loss: 6.045 I0420 20:40:59.904426 135309809014592 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0420 20:41:00.195712 135309809014592 checkpointing.py:798] Waited 0.29121994972229004 seconds for step 15 to finish before starting checkpointing. I0420 20:41:00.198522 135309809014592 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 20:41:03.768702 135186847102720 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/10/items/array_metadatas/process_7 I0420 20:41:41.301389 135192316458752 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 37.015 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.67506527900696 s) (per-host) I0420 20:41:41.301513 135192316458752 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.132775s. I0420 20:41:48.116216 135192316458752 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 48.947462s. I0420 20:41:48.116514 135186830317312 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 20:41:48.116631 135186830317312 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 20:41:48.116693 135186830317312 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 20:41:48.118341 135186830317312 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 20:41:48.118509 135309809014592 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0420 20:41:48.118640 135309809014592 checkpoint_manager.py:1458] Waiting for previous save to complete took 47.920121 seconds. If this number is high, consider checkpointing less frequently. I0420 20:41:48.120688 135309809014592 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 15 I0420 20:41:48.124727 135309809014592 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/15. I0420 20:41:48.696376 135309809014592 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0420 20:41:48.696475 135309809014592 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 I0420 20:41:48.743360 135309809014592 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.048279s I0420 20:41:48.743526 135309809014592 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.781 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3226161003112793 s) (per-host) I0420 20:41:48.743574 135309809014592 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.322673s (batch_requests_ready=0.272446s, total_serialization_initiated=0.050163s, others=0.000064s) I0420 20:41:48.743675 135309809014592 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.327630s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.327618) I0420 20:41:48.744541 135188013098752 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 20:41:48.744731 135309809014592 async_checkpointer.py:561] Finished blocking save. Time taken: 0.623972s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/15. I0420 20:41:48.746647 135309809014592 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0420 20:41:48.746937 135186830317312 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 20:41:48.747097 135309809014592 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776717660.1984928, 'wait_for_prev_duration_secs': 47.92012119293213, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776717708.1207254, 'checkpointer_blocking_duration_secs': 0.624114990234375, 'get_old_steps_start_time': 1776717708.744859, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776717660.1964173, 'checkpoint_manager_blocking_duration_secs': 48.55064678192139} I0420 20:41:48.747200 135309809014592 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0420 20:41:48.747758 135309809014592 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.950, Tokens/s/device: 13859.657, total_weights: 65536, loss: 5.864 I0420 20:41:48.753067 135309809014592 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 92.063, Tokens/s/device: 13876.749, total_weights: 65536, loss: 5.719 I0420 20:42:06.452535 135309809014592 metric_logger.py:185] completed step: 16, seconds: 48.847, TFLOP/s/device: 0.278, Tokens/s/device: 41.926, total_weights: 65536, loss: 5.605 I0420 20:42:06.600099 135309809014592 metric_logger.py:185] completed step: 17, seconds: 17.553, TFLOP/s/device: 0.774, Tokens/s/device: 116.673, total_weights: 65536, loss: 5.517 I0420 20:42:06.747459 135309809014592 metric_logger.py:185] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.437, Tokens/s/device: 13480.957, total_weights: 65536, loss: 5.448 I0420 20:42:06.894646 135309809014592 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 20:42:06.895350 135309809014592 checkpointing.py:798] Waited 0.0007233619689941406 seconds for step 19 to finish before starting checkpointing. I0420 20:42:06.897268 135309809014592 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 20:42:10.913580 135186847102720 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/15/items/array_metadatas/process_7 I0420 20:42:47.671540 135188013098752 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 26.660 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 59.25060677528381 s) (per-host) I0420 20:42:47.671644 135188013098752 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 58.926994s. I0420 20:42:55.822354 135188013098752 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 67.077687s. I0420 20:42:55.822708 135186830317312 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 20:42:55.822832 135186830317312 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 20:42:55.822881 135186830317312 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0420 20:42:55.825383 135186830317312 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0420 20:42:55.825598 135309809014592 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0420 20:42:55.825734 135309809014592 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.928477 seconds. If this number is high, consider checkpointing less frequently. I0420 20:42:55.827385 135309809014592 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 19 I0420 20:42:55.830926 135309809014592 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/19. I0420 20:42:56.410792 135309809014592 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0420 20:42:56.410887 135309809014592 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 I0420 20:42:56.458225 135309809014592 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.048342s I0420 20:42:56.458390 135309809014592 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.923 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3133540153503418 s) (per-host) I0420 20:42:56.458439 135309809014592 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.313413s (batch_requests_ready=0.262896s, total_serialization_initiated=0.050453s, others=0.000064s) I0420 20:42:56.458528 135309809014592 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.317422s (all_items=0.000010s, per_item={'items': '0.00001001'}, temp_paths=0.317412) I0420 20:42:56.459453 135192308066048 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 20:42:56.459614 135309809014592 async_checkpointer.py:561] Finished blocking save. Time taken: 0.632160s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/19. I0420 20:42:56.461624 135309809014592 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 20:42:56.461924 135186830317312 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 20:42:56.462062 135309809014592 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776717726.8972402, 'wait_for_prev_duration_secs': 48.9284770488739, 'time_between_consecutive_saves_sec': 18.778858423233032, 'checkpointer_blocking_start_time': 1776717775.8274238, 'checkpointer_blocking_duration_secs': 0.6323175430297852, 'get_old_steps_start_time': 1776717776.459755, 'get_old_steps_duration_secs': 2.1219253540039062e-05, 'checkpoint_manager_blocking_start_time': 1776717726.8955765, 'checkpoint_manager_blocking_duration_secs': 49.566455364227295} I0420 20:42:56.462217 135309809014592 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 20:42:56.462268 135309809014592 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 20:43:01.236004 135192316458752 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_16_zarr3_true/checkpoints/19/items/array_metadatas/process_7 I0420 20:43:37.667251 135192308066048 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 38.042 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.52217125892639 s) (per-host) I0420 20:43:37.667369 135192308066048 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.207803s. I0420 20:43:44.883306 135192308066048 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 48.423726s. I0420 20:43:44.883587 135186830317312 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 20:43:44.883736 135186830317312 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 20:43:44.883778 135186830317312 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 20:43:44.885424 135186830317312 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 20:43:44.885612 135309809014592 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0420 20:43:44.885774 135309809014592 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 20:43:44.886460 135309809014592 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.161, Tokens/s/device: 13891.527, 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: Mon Apr 20 20:43:55 UTC 2026 EXIT_CODE=0