XPK Start: Fri Apr 24 16:59:22 UTC 2026 2026-04-24 16:59:26.846243: 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 00:00:1777049966.859349 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777049966.863087 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777049966.874785 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777049966.874803 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777049966.874806 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777049966.874808 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 16:59:46.090184: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0424 16:59:46.619284 132423391360832 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 16:59:55,660:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 16:59:55.660289 132423391360832 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 16:59:55,662:jax._src.distributed:157: Connecting to JAX distributed service on mt-09-pdb-lt-1-k4gje-slice-job-0-0.mt-09-pdb-lt-1-k4gje:8482 I0424 16:59:55.662619 132423391360832 distributed.py:157] Connecting to JAX distributed service on mt-09-pdb-lt-1-k4gje-slice-job-0-0.mt-09-pdb-lt-1-k4gje:8482 I0424 16:59:56.661879 132423391360832 max_utils.py:284] Jax distributed system initialized! I0424 17:00:03.817690 132423391360832 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 17:00:03.817795 132423391360832 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 17:00:03.817842 132423391360832 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0424 17:00:03.817889 132423391360832 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 17:00:04.443283 132423391360832 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0424 17:00:04.557215 132423391360832 checkpointing.py:688] Setting up checkpoint logger... I0424 17:00:04.557343 132423391360832 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 17:00:04.557388 132423391360832 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 17:00:04.557597 132423391360832 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 0x786fa5d7d280>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 17:00:07.484000 132423391360832 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 17:00:07.484262 132423391360832 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 0x785b1466eea0>}, handler_registry=None I0424 17:00:07.484505 132423391360832 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x785b1466eea0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0424 17:00:07.484554 132423391360832 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x785b1466c170>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0424 17:00:07.484589 132423391360832 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 0x785b1466eea0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x785b1466eea0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x785b1466c170>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x785b1466c170>}). I0424 17:00:07.484919 132423391360832 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 17:00:07.484995 132423391360832 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x785b14418f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 17:00:13.921074 132423391360832 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints I0424 17:00:14.049405 132423391360832 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_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x785b1466c320> I0424 17:00:14.049583 132423391360832 checkpointing.py:302] Checkpoint manager created! I0424 17:00:14.993294 132423391360832 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 17:00:14.993414 132423391360832 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0424 17:00:16.416672 132423391360832 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: ('tensor',) I0424 17:00:16.416796 132423391360832 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0424 17:00:16.416850 132423391360832 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0424 17:00:16.416895 132423391360832 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: ('tensor', None, 'fsdp') I0424 17:00:16.416932 132423391360832 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0424 17:00:16.416969 132423391360832 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0424 17:00:16.417007 132423391360832 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 17:00:16.417044 132423391360832 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: ('tensor', None, None, 'fsdp') I0424 17:00:16.417080 132423391360832 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 17:00:16.417127 132423391360832 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 17:00:16.417164 132423391360832 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', 'tensor') I0424 17:00:16.417198 132423391360832 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: ('tensor', 'fsdp') I0424 17:00:16.814715 132423391360832 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:00:16.814865 132423391360832 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 17:00:16.820325 132423391360832 nnx_decoders.py:616] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 17:00:16.820424 132423391360832 nnx_decoders.py:616] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 17:00:16.975989 132423391360832 attentions.py:1084] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:00:16.976142 132423391360832 attentions.py:1084] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 17:00:17.130053 132423391360832 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 17:00:17.130207 132423391360832 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 17:00:17.294328 132423391360832 attentions.py:1150] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:00:17.294464 132423391360832 attentions.py:1150] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 17:00:17.453121 132423391360832 attentions.py:1151] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:00:17.453264 132423391360832 attentions.py:1151] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 17:00:17.610546 132423391360832 attentions.py:1152] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 17:00:17.610682 132423391360832 attentions.py:1152] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 17:00:17.776938 132423391360832 attentions.py:1193] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 17:00:17.777079 132423391360832 attentions.py:1193] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 17:00:17.800158 132423391360832 linears.py:541] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 17:00:17.800264 132423391360832 linears.py:541] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0424 17:00:28.156833 132423391360832 max_utils.py:791] Total memory size: 0.9 GB, Output size: 0.4 GB, Temp size: 0.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0424 17:00:28.159223 132423391360832 metric_logger.py:289] number parameters: 1.104 billion I0424 17:00:29.477965 132423391360832 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 17:00:37.760701 132423391360832 checkpointing.py:798] Waited 8.282718658447266 seconds for step 0 to finish before starting checkpointing. I0424 17:00:37.762989 132423391360832 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:00:37.764781 132423391360832 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0424 17:00:37.767470 132423391360832 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/0. I0424 17:00:38.388054 132423391360832 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 17:00:38.389231 132423391360832 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:00:38.389293 132423391360832 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0424 17:00:38.620472 132423391360832 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.233551s I0424 17:00:38.620658 132423391360832 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.955 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5220293998718262 s) (per-host) I0424 17:00:38.620713 132423391360832 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.522095s (batch_requests_ready=0.282058s, total_serialization_initiated=0.239964s, others=0.000073s) I0424 17:00:38.620814 132423391360832 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.526195s (all_items=0.000026s, per_item={'items': '0.00002623'}, temp_paths=0.526169) I0424 17:00:38.621799 132300385789696 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:00:38.621973 132423391360832 async_checkpointer.py:561] Finished blocking save. Time taken: 0.857121s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/0. I0424 17:00:38.661895 132423391360832 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:00:38.662256 132297189742336 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:00:38.662417 132423391360832 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777050037.7629702, 'wait_for_prev_duration_secs': 6.508827209472656e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777050037.7648208, 'checkpointer_blocking_duration_secs': 0.8572578430175781, 'get_old_steps_start_time': 1777050038.6221135, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1777050037.7611213, 'checkpoint_manager_blocking_duration_secs': 0.9012560844421387} I0424 17:00:38.662556 132423391360832 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 17:00:38.662638 132423391360832 max_utils.py:750] Memstats: After params initialized: I0424 17:00:38.662692 132423391360832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 17:00:38.662726 132423391360832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 17:00:38.662754 132423391360832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 17:00:38.662778 132423391360832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 17:00:38.935867 132423391360832 metric_logger.py:185] completed step: 0, seconds: 1.318, TFLOP/s/device: 2.578, Tokens/s/device: 388.563, total_weights: 16384, loss: 10.864 I0424 17:00:39.111510 132423391360832 metric_logger.py:185] completed step: 1, seconds: 9.456, TFLOP/s/device: 0.359, Tokens/s/device: 54.144, total_weights: 16384, loss: 10.864 I0424 17:00:39.565035 132423391360832 metric_logger.py:185] completed step: 2, seconds: 0.121, TFLOP/s/device: 28.188, Tokens/s/device: 4248.822, total_weights: 16384, loss: 9.760 I0424 17:00:39.623378 132423391360832 metric_logger.py:185] completed step: 3, seconds: 0.442, TFLOP/s/device: 7.677, Tokens/s/device: 1157.107, total_weights: 16384, loss: 8.820 I0424 17:00:39.741349 132423391360832 metric_logger.py:185] completed step: 4, seconds: 0.074, TFLOP/s/device: 45.955, Tokens/s/device: 6926.782, total_weights: 16384, loss: 7.947 I0424 17:00:39.748502 132423391360832 metric_logger.py:185] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.919, Tokens/s/device: 8880.871, total_weights: 16384, loss: 7.209 I0424 17:00:52.079226 132423391360832 metric_logger.py:185] completed step: 6, seconds: 0.119, TFLOP/s/device: 28.588, Tokens/s/device: 4309.111, total_weights: 16384, loss: 6.651 I0424 17:00:52.137841 132423391360832 metric_logger.py:185] completed step: 7, seconds: 12.273, TFLOP/s/device: 0.277, Tokens/s/device: 41.716, total_weights: 16384, loss: 6.267 I0424 17:00:52.196354 132423391360832 metric_logger.py:185] completed step: 8, seconds: 0.064, TFLOP/s/device: 52.755, Tokens/s/device: 7951.792, total_weights: 16384, loss: 6.030 I0424 17:00:52.255824 132423391360832 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 17:00:52.256418 132423391360832 checkpointing.py:798] Waited 0.0006046295166015625 seconds for step 9 to finish before starting checkpointing. I0424 17:00:52.258477 132423391360832 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:00:53.673381 2422 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 17:00:55.613743 132300394182400 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_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_6 I0424 17:01:27.741379 132300385789696 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 31.819 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 49.64271569252014 s) (per-host) I0424 17:01:27.741481 132300385789696 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.119567s. I0424 17:01:34.277963 132300385789696 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 55.656034s. I0424 17:01:34.278280 132297189742336 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:01:34.278404 132297189742336 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:01:34.278455 132297189742336 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:01:34.280033 132297189742336 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 17:01:34.280213 132423391360832 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. W0424 17:01:34.280356 132423391360832 checkpoint_manager.py:1458] Waiting for previous save to complete took 42.021884 seconds. If this number is high, consider checkpointing less frequently. I0424 17:01:34.282701 132423391360832 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0424 17:01:34.286091 132423391360832 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/9. I0424 17:01:34.838221 132423391360832 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0424 17:01:34.838375 132423391360832 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0424 17:01:34.886961 132423391360832 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.050755s I0424 17:01:34.887141 132423391360832 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.781 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.32262206077575684 s) (per-host) I0424 17:01:34.887196 132423391360832 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.322686s (batch_requests_ready=0.266957s, total_serialization_initiated=0.055659s, others=0.000071s) I0424 17:01:34.887295 132423391360832 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.326797s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=0.326783) I0424 17:01:34.888159 132297172956928 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0424 17:01:34.888317 132423391360832 async_checkpointer.py:561] Finished blocking save. Time taken: 0.605540s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/9. I0424 17:01:34.890183 132423391360832 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 17:01:34.890469 132297189742336 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 17:01:34.890604 132423391360832 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777050052.2584472, 'wait_for_prev_duration_secs': 42.02188444137573, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777050094.2827451, 'checkpointer_blocking_duration_secs': 0.6056790351867676, 'get_old_steps_start_time': 1777050094.888443, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1777050052.2566502, 'checkpoint_manager_blocking_duration_secs': 42.633920192718506} I0424 17:01:34.890765 132423391360832 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 17:01:34.890812 132423391360832 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 17:01:40.063081 132300394182400 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_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_6 I0424 17:02:19.285447 132297172956928 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 35.321 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.72088551521301 s) (per-host) I0424 17:02:19.285571 132297172956928 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.397302s. I0424 17:02:23.516643 132297172956928 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 48.628358s. I0424 17:02:23.516945 132297189742336 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 17:02:23.517066 132297189742336 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 17:02:23.517135 132297189742336 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 17:02:23.519298 132297189742336 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 17:02:23.519475 132423391360832 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. I0424 17:02:23.519643 132423391360832 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 17:02:23.520289 132423391360832 metric_logger.py:185] completed step: 9, seconds: 0.058, TFLOP/s/device: 58.635, Tokens/s/device: 8838.100, total_weights: 16384, loss: 5.890 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 17:02:33 UTC 2026 EXIT_CODE=0