XPK Start: Thu Apr 23 17:33:54 UTC 2026 2026-04-23 17:33:58.948624: 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 00:00:1776965638.961688 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776965638.965482 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776965638.977198 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965638.977217 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965638.977219 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776965638.977221 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 17:34:18.165558: 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) I0423 17:34:18.694601 139099505485632 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 17:34:27,735:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 17:34:27.735390 139099505485632 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 17:34:27,737:jax._src.distributed:157: Connecting to JAX distributed service on mt-01-base-3scfz-slice-job-0-0.mt-01-base-3scfz:8482 I0423 17:34:27.737704 139099505485632 distributed.py:157] Connecting to JAX distributed service on mt-01-base-3scfz-slice-job-0-0.mt-01-base-3scfz:8482 I0423 17:34:28.911990 139099505485632 max_utils.py:284] Jax distributed system initialized! I0423 17:34:35.955346 139099505485632 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 17:34:35.955455 139099505485632 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 17:34:35.955496 139099505485632 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 I0423 17:34:35.955536 139099505485632 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 17:34:36.582808 139099505485632 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 17:34:36.699064 139099505485632 checkpointing.py:688] Setting up checkpoint logger... I0423 17:34:36.699198 139099505485632 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 17:34:36.699247 139099505485632 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 17:34:36.699511 139099505485632 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 0x7e82381a5e20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 17:34:40.470873 139099505485632 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 17:34:40.471137 139099505485632 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 0x7e6fba1f6210>}, handler_registry=None I0423 17:34:40.471385 139099505485632 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6fba1f6210>` 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`. I0423 17:34:40.471436 139099505485632 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6fb9d72d20>` 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`. I0423 17:34:40.471474 139099505485632 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 0x7e6fba1f6210>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6fba1f6210>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6fb9d72d20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6fb9d72d20>}). I0423 17:34:40.471830 139099505485632 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 17:34:40.471908 139099505485632 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e6fb8ed8f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 17:34:41.556504 139099505485632 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints I0423 17:34:41.858765 139099505485632 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=10), 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e6fba6ebcb0> I0423 17:34:41.858943 139099505485632 checkpointing.py:302] Checkpoint manager created! I0423 17:34:42.784877 139099505485632 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 17:34:42.784998 139099505485632 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0423 17:34:44.230536 139099505485632 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 17:34:44.230667 139099505485632 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:34:44.230725 139099505485632 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:34:44.230770 139099505485632 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 17:34:44.230812 139099505485632 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:34:44.230849 139099505485632 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:34:44.230889 139099505485632 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:34:44.230928 139099505485632 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 17:34:44.230964 139099505485632 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:34:44.230999 139099505485632 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:34:44.231034 139099505485632 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 17:34:44.231070 139099505485632 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 17:34:44.622999 139099505485632 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:34:44.623170 139099505485632 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:34:44.628571 139099505485632 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:34:44.628669 139099505485632 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:34:44.784731 139099505485632 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:34:44.784872 139099505485632 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:34:44.938953 139099505485632 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:34:44.939116 139099505485632 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:34:45.104154 139099505485632 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:34:45.104295 139099505485632 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:34:45.256805 139099505485632 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:34:45.256945 139099505485632 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:34:45.406775 139099505485632 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:34:45.406912 139099505485632 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:34:45.568750 139099505485632 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 17:34:45.568902 139099505485632 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:34:45.591598 139099505485632 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 17:34:45.591707 139099505485632 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 17:34:56.813914 139099505485632 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. I0423 17:34:56.816311 139099505485632 metric_logger.py:289] number parameters: 1.104 billion I0423 17:35:07.916270 139099505485632 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0423 17:35:08.068018 139099505485632 checkpointing.py:798] Waited 0.1517329216003418 seconds for step 0 to finish before starting checkpointing. I0423 17:35:08.070321 139099505485632 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:35:08.071993 139099505485632 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0423 17:35:08.074883 139099505485632 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/0. I0423 17:35:08.932283 139099505485632 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 17:35:08.933416 139099505485632 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 17:35:08.933475 139099505485632 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 I0423 17:35:09.181700 139099505485632 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.250599s I0423 17:35:09.181876 139099505485632 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.965 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7851846218109131 s) (per-host) I0423 17:35:09.181929 139099505485632 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.785248s (batch_requests_ready=0.527957s, total_serialization_initiated=0.257220s, others=0.000071s) I0423 17:35:09.182027 139099505485632 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.789392s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.789372) I0423 17:35:09.182985 138981542962944 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:35:09.183184 139099505485632 async_checkpointer.py:561] Finished blocking save. Time taken: 1.111118s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/0. I0423 17:35:09.185080 139099505485632 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:35:09.185402 138977776473856 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:35:09.185532 139099505485632 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776965708.0703034, 'wait_for_prev_duration_secs': 6.651878356933594e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776965708.072035, 'checkpointer_blocking_duration_secs': 1.1112606525421143, 'get_old_steps_start_time': 1776965709.1833236, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776965708.0684454, 'checkpoint_manager_blocking_duration_secs': 1.1170494556427002} I0423 17:35:09.185698 139099505485632 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0423 17:35:09.185773 139099505485632 max_utils.py:750] Memstats: After params initialized: I0423 17:35:09.185825 139099505485632 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 17:35:09.185858 139099505485632 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 17:35:09.185885 139099505485632 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 17:35:09.185912 139099505485632 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 17:35:09.458628 139099505485632 metric_logger.py:185] completed step: 0, seconds: 11.099, TFLOP/s/device: 1.224, Tokens/s/device: 184.523, total_weights: 65536, loss: 10.874 I0423 17:35:09.625339 139099505485632 metric_logger.py:185] completed step: 1, seconds: 1.541, TFLOP/s/device: 8.819, Tokens/s/device: 1329.343, total_weights: 65536, loss: 10.874 I0423 17:35:10.053299 139099505485632 metric_logger.py:185] completed step: 2, seconds: 0.021, TFLOP/s/device: 646.852, Tokens/s/device: 97500.595, total_weights: 65536, loss: 10.262 I0423 17:35:10.200718 139099505485632 metric_logger.py:185] completed step: 3, seconds: 0.419, TFLOP/s/device: 32.419, Tokens/s/device: 4886.487, total_weights: 65536, loss: 9.732 I0423 17:35:10.496144 139099505485632 metric_logger.py:185] completed step: 4, seconds: 0.163, TFLOP/s/device: 83.462, Tokens/s/device: 12580.316, total_weights: 65536, loss: 9.274 I0423 17:35:10.503621 139099505485632 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.547, Tokens/s/device: 13949.718, total_weights: 65536, loss: 8.888 I0423 17:35:29.241759 139099505485632 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.898, Tokens/s/device: 6918.218, total_weights: 65536, loss: 8.589 I0423 17:35:29.388991 139099505485632 metric_logger.py:185] completed step: 7, seconds: 18.592, TFLOP/s/device: 0.731, Tokens/s/device: 110.154, total_weights: 65536, loss: 8.380 I0423 17:35:29.536542 139099505485632 metric_logger.py:185] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.637, Tokens/s/device: 13360.297, total_weights: 65536, loss: 8.251 I0423 17:35:29.684984 139099505485632 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0423 17:35:29.685600 139099505485632 checkpointing.py:798] Waited 0.0006263256072998047 seconds for step 9 to finish before starting checkpointing. I0423 17:35:29.687406 139099505485632 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:35:30.417202 2549 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 17:35:32.656494 138981551355648 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/0/items/array_metadatas/process_6 I0423 17:36:03.796954 138981542962944 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 28.512 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 55.40022325515747 s) (per-host) I0423 17:36:03.797062 138981542962944 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 54.613942s. I0423 17:36:12.146614 138981542962944 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 62.963476s. I0423 17:36:12.146906 138977776473856 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:36:12.147034 138977776473856 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:36:12.147084 138977776473856 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:36:12.148582 138977776473856 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 17:36:12.148777 139099505485632 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. W0423 17:36:12.148914 139099505485632 checkpoint_manager.py:1458] Waiting for previous save to complete took 42.461512 seconds. If this number is high, consider checkpointing less frequently. I0423 17:36:12.150708 139099505485632 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0423 17:36:12.154224 139099505485632 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/9. I0423 17:36:12.695400 139099505485632 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0423 17:36:12.695559 139099505485632 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 I0423 17:36:12.736656 139099505485632 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043212s I0423 17:36:12.736841 139099505485632 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.096 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30273008346557617 s) (per-host) I0423 17:36:12.736908 139099505485632 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.302805s (batch_requests_ready=0.254787s, total_serialization_initiated=0.047933s, others=0.000084s) I0423 17:36:12.737028 139099505485632 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.306748s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.306733) I0423 17:36:12.737906 138977818437376 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:36:12.738066 139099505485632 async_checkpointer.py:561] Finished blocking save. Time taken: 0.587283s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/9. I0423 17:36:12.752452 139099505485632 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:36:12.752763 138977776473856 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:36:12.752964 139099505485632 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776965729.687378, 'wait_for_prev_duration_secs': 42.461512327194214, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776965772.1507487, 'checkpointer_blocking_duration_secs': 0.5874390602111816, 'get_old_steps_start_time': 1776965772.738205, 'get_old_steps_duration_secs': 2.4080276489257812e-05, 'checkpoint_manager_blocking_start_time': 1776965729.68583, 'checkpoint_manager_blocking_duration_secs': 43.067100048065186} I0423 17:36:12.753073 139099505485632 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0423 17:36:12.753132 139099505485632 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:36:17.666264 138978328942336 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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_01_base/checkpoints/9/items/array_metadatas/process_6 I0423 17:36:54.374895 138977818437376 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 37.663 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.94074821472168 s) (per-host) I0423 17:36:54.375013 138977818437376 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.636996s. I0423 17:37:02.146465 138977818437376 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 49.408431s. I0423 17:37:02.146788 138977776473856 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:37:02.146907 138977776473856 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:37:02.146955 138977776473856 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:37:02.148627 138977776473856 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 17:37:02.148826 139099505485632 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. I0423 17:37:02.148994 139099505485632 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:37:02.149791 139099505485632 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.426, Tokens/s/device: 13931.404, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 23 17:37:15 UTC 2026 EXIT_CODE=0