XPK Start: Wed Apr 22 18:39:02 UTC 2026 2026-04-22 18:39:06.554140: 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 00:00:1776883146.567180 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776883146.571106 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776883146.582975 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883146.582993 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883146.582995 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883146.582997 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-22 18:39:25.818083: 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 18:39:26.343616 135050481198912 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-22 18:39:35,384:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0422 18:39:35.384270 135050481198912 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-22 18:39:35,386:jax._src.distributed:157: Connecting to JAX distributed service on mt-15-ocdbt-false-sv-re7tk-slice-job-0-0.mt-15-ocdbt-false-sv-re7tk:8482 I0422 18:39:35.386523 135050481198912 distributed.py:157] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-re7tk-slice-job-0-0.mt-15-ocdbt-false-sv-re7tk:8482 I0422 18:39:36.361297 135050481198912 max_utils.py:284] Jax distributed system initialized! I0422 18:39:43.239832 135050481198912 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:39:43.239939 135050481198912 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:39:43.239980 135050481198912 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 18:39:43.240015 135050481198912 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:39:43.857212 135050481198912 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:39:43.971570 135050481198912 checkpointing.py:688] Setting up checkpoint logger... I0422 18:39:43.971685 135050481198912 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0422 18:39:43.971732 135050481198912 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0422 18:39:43.971936 135050481198912 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 0x7ad3bfb0cfe0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:39:46.859053 135050481198912 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 18:39:46.859442 135050481198912 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 0x7ad21f1ee660>}, handler_registry=None I0422 18:39:46.859694 135050481198912 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ad21f1ee660>` 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 18:39:46.859745 135050481198912 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7abeb82604a0>` 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 18:39:46.859781 135050481198912 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 0x7ad21f1ee660>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ad21f1ee660>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7abeb82604a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7abeb82604a0>}). I0422 18:39:46.860126 135050481198912 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0422 18:39:46.860205 135050481198912 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7abee8234f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:39:48.169183 135050481198912 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints I0422 18:39:48.607233 135050481198912 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/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7abee82aac00> I0422 18:39:48.607416 135050481198912 checkpointing.py:302] Checkpoint manager created! I0422 18:39:49.555635 135050481198912 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:39:49.555757 135050481198912 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0422 18:39:51.068131 135050481198912 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0422 18:39:51.068272 135050481198912 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 18:39:51.068347 135050481198912 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 18:39:51.068411 135050481198912 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0422 18:39:51.068469 135050481198912 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 18:39:51.068532 135050481198912 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 18:39:51.068602 135050481198912 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 18:39:51.068673 135050481198912 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0422 18:39:51.068743 135050481198912 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 18:39:51.068807 135050481198912 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 18:39:51.068864 135050481198912 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0422 18:39:51.068906 135050481198912 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0422 18:39:51.472947 135050481198912 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:39:51.473113 135050481198912 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 18:39:51.478716 135050481198912 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:39:51.478809 135050481198912 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 18:39:51.626121 135050481198912 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:39:51.626254 135050481198912 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 18:39:51.774890 135050481198912 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:39:51.775023 135050481198912 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 18:39:51.930916 135050481198912 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 18:39:51.931053 135050481198912 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 18:39:52.086209 135050481198912 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 18:39:52.086343 135050481198912 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 18:39:52.234730 135050481198912 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 18:39:52.234862 135050481198912 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 18:39:52.394335 135050481198912 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:39:52.394475 135050481198912 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 18:39:52.418212 135050481198912 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:39:52.418315 135050481198912 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 18:40:03.287152 135050481198912 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 18:40:03.291051 135050481198912 metric_logger.py:289] number parameters: 1.104 billion I0422 18:40:14.457077 135050481198912 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0422 18:40:14.607625 135050481198912 checkpointing.py:798] Waited 0.15053343772888184 seconds for step 0 to finish before starting checkpointing. I0422 18:40:14.609941 135050481198912 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:40:14.611721 135050481198912 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0422 18:40:14.614490 135050481198912 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/0. I0422 18:40:15.625767 135050481198912 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 18:40:15.626796 135050481198912 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0422 18:40:15.626855 135050481198912 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 18:40:15.874207 135050481198912 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.249578s I0422 18:40:15.874377 135050481198912 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.004 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5134782791137695 s) (per-host) I0422 18:40:15.874431 135050481198912 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.513542s (batch_requests_ready=0.257308s, total_serialization_initiated=0.256163s, others=0.000071s) I0422 18:40:15.874529 135050481198912 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.517517s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.517497) I0422 18:40:15.875501 134927462950656 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0422 18:40:15.875669 135050481198912 async_checkpointer.py:561] Finished blocking save. Time taken: 1.263875s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/0. I0422 18:40:16.332943 135050481198912 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:40:16.333330 134922622719744 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:40:16.333497 135050481198912 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883214.6099231, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883214.6117606, 'checkpointer_blocking_duration_secs': 1.2640137672424316, 'get_old_steps_start_time': 1776883215.875793, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776883214.6080248, 'checkpoint_manager_blocking_duration_secs': 1.7254314422607422} I0422 18:40:16.333607 135050481198912 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0422 18:40:16.333680 135050481198912 max_utils.py:750] Memstats: After params initialized: I0422 18:40:16.333735 135050481198912 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0422 18:40:16.333770 135050481198912 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0422 18:40:16.333795 135050481198912 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0422 18:40:16.333819 135050481198912 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0422 18:40:16.602363 135050481198912 metric_logger.py:185] completed step: 0, seconds: 11.165, TFLOP/s/device: 1.217, Tokens/s/device: 183.431, total_weights: 65536, loss: 10.874 I0422 18:40:16.769796 135050481198912 metric_logger.py:185] completed step: 1, seconds: 2.144, TFLOP/s/device: 6.339, Tokens/s/device: 955.434, total_weights: 65536, loss: 10.874 I0422 18:40:17.197932 135050481198912 metric_logger.py:185] completed step: 2, seconds: 0.022, TFLOP/s/device: 618.131, Tokens/s/device: 93171.375, total_weights: 65536, loss: 10.560 I0422 18:40:17.345264 135050481198912 metric_logger.py:185] completed step: 3, seconds: 0.419, TFLOP/s/device: 32.433, Tokens/s/device: 4888.622, total_weights: 65536, loss: 9.979 I0422 18:40:17.351415 135050481198912 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0422 18:40:17.640348 135050481198912 checkpointing.py:798] Waited 0.2889058589935303 seconds for step 5 to finish before starting checkpointing. I0422 18:40:17.643077 135050481198912 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:40:18.559224 2652 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 18:40:20.730011 134927471343360 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_6 I0422 18:40:26.302654 134927462950656 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 144.365 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.941715002059937 s) (per-host) I0422 18:40:26.302769 134927462950656 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.427156s. I0422 18:40:33.420499 134927462950656 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 17.544871s. I0422 18:40:33.420734 134922622719744 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:40:33.420792 134922622719744 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:40:33.420841 134922622719744 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:40:33.422283 134922622719744 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 18:40:33.422460 135050481198912 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0422 18:40:33.422597 135050481198912 checkpoint_manager.py:1458] Waiting for previous save to complete took 15.779523 seconds. If this number is high, consider checkpointing less frequently. I0422 18:40:33.424656 135050481198912 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 5 I0422 18:40:33.428937 135050481198912 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/5. I0422 18:40:33.972075 135050481198912 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0422 18:40:33.972240 135050481198912 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 18:40:34.016889 135050481198912 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046877s I0422 18:40:34.017060 135050481198912 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.998 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3086674213409424 s) (per-host) I0422 18:40:34.017136 135050481198912 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.308748s (batch_requests_ready=0.256597s, total_serialization_initiated=0.052062s, others=0.000089s) I0422 18:40:34.017245 135050481198912 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.313674s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.313659) I0422 18:40:34.018157 134918310967040 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0422 18:40:34.018323 135050481198912 async_checkpointer.py:561] Finished blocking save. Time taken: 0.593596s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/5. I0422 18:40:34.023604 135050481198912 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:40:34.023831 134922622719744 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:40:34.023960 135050481198912 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883217.6430488, 'wait_for_prev_duration_secs': 15.779523372650146, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883233.4246957, 'checkpointer_blocking_duration_secs': 0.5937361717224121, 'get_old_steps_start_time': 1776883234.0184507, 'get_old_steps_duration_secs': 2.4557113647460938e-05, 'checkpoint_manager_blocking_start_time': 1776883217.6407044, 'checkpoint_manager_blocking_duration_secs': 16.38322114944458} I0422 18:40:34.024067 135050481198912 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0422 18:40:34.024831 135050481198912 metric_logger.py:185] completed step: 4, seconds: 0.163, TFLOP/s/device: 83.524, Tokens/s/device: 12589.674, total_weights: 65536, loss: 9.457 I0422 18:40:34.032114 135050481198912 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.335, Tokens/s/device: 13917.676, total_weights: 65536, loss: 8.955 I0422 18:40:53.790443 135050481198912 metric_logger.py:185] completed step: 6, seconds: 16.679, TFLOP/s/device: 0.815, Tokens/s/device: 122.789, total_weights: 65536, loss: 8.465 I0422 18:40:53.937808 135050481198912 metric_logger.py:185] completed step: 7, seconds: 19.613, TFLOP/s/device: 0.693, Tokens/s/device: 104.421, total_weights: 65536, loss: 7.998 I0422 18:40:54.085327 135050481198912 metric_logger.py:185] completed step: 8, seconds: 0.154, TFLOP/s/device: 88.476, Tokens/s/device: 13336.112, total_weights: 65536, loss: 7.567 I0422 18:40:54.233890 135050481198912 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0422 18:40:54.234570 135050481198912 checkpointing.py:798] Waited 0.0006833076477050781 seconds for step 9 to finish before starting checkpointing. I0422 18:40:54.236377 135050481198912 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:40:58.397448 134927471343360 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_6 I0422 18:41:04.189052 134918310967040 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 51.823 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 30.48060154914856 s) (per-host) I0422 18:41:04.189188 134918310967040 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.170911s. I0422 18:41:10.903502 134918310967040 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 36.885210s. I0422 18:41:10.903710 134922622719744 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:41:10.903772 134922622719744 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:41:10.903819 134922622719744 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:41:10.906514 134922622719744 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0422 18:41:10.906708 135050481198912 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0422 18:41:10.906832 135050481198912 checkpoint_manager.py:1458] Waiting for previous save to complete took 16.670467 seconds. If this number is high, consider checkpointing less frequently. I0422 18:41:10.908416 135050481198912 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0422 18:41:10.911964 135050481198912 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/9. I0422 18:41:11.456079 135050481198912 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0422 18:41:11.456193 135050481198912 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 18:41:11.499009 135050481198912 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045028s I0422 18:41:11.499199 135050481198912 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.126 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3009335994720459 s) (per-host) I0422 18:41:11.499250 135050481198912 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.300993s (batch_requests_ready=0.251185s, total_serialization_initiated=0.049744s, others=0.000064s) I0422 18:41:11.499360 135050481198912 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.305286s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.305275) I0422 18:41:11.500308 134927462950656 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0422 18:41:11.500472 135050481198912 async_checkpointer.py:561] Finished blocking save. Time taken: 0.591981s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/9. I0422 18:41:11.930065 135050481198912 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:41:11.930452 134922622719744 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:41:11.930627 135050481198912 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883254.236348, 'wait_for_prev_duration_secs': 16.670467138290405, 'time_between_consecutive_saves_sec': 20.814025163650513, 'checkpointer_blocking_start_time': 1776883270.9084566, 'checkpointer_blocking_duration_secs': 0.5921401977539062, 'get_old_steps_start_time': 1776883271.500613, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1776883254.234792, 'checkpoint_manager_blocking_duration_secs': 17.695800065994263} I0422 18:41:11.930748 135050481198912 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0422 18:41:11.930790 135050481198912 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:41:16.655179 134927471343360 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260422_154647/nnx_xpk_feat_nnx_set_defaults_true_20260422_154647_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_6 I0422 18:41:22.243848 134927462950656 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 143.008 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.045543909072876 s) (per-host) I0422 18:41:22.243963 134927462950656 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.743543s. I0422 18:41:27.592868 134927462950656 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 16.092433s. I0422 18:41:27.593191 134922622719744 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:41:27.593312 134922622719744 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:41:27.593362 134922622719744 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:41:27.595879 134922622719744 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 18:41:27.596070 135050481198912 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0422 18:41:27.596282 135050481198912 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:41:27.596894 135050481198912 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.485, Tokens/s/device: 13940.318, total_weights: 65536, loss: 7.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Wed Apr 22 18:41:39 UTC 2026 EXIT_CODE=0