XPK Start: Sat Apr 25 17:43:41 UTC 2026 2026-04-25 17:43:45.293553: 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 00:00:1777139025.306774 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777139025.310772 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777139025.322268 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777139025.322287 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777139025.322289 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777139025.322291 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-25 17:44:04.550772: 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) I0425 17:44:05.083820 134080978020160 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-25 17:44:14,126:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0425 17:44:14.126868 134080978020160 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-25 17:44:14,129:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-true-sv-b6fg5-slice-job-0-0.mt-16-zarr3-true-sv-b6fg5:8482 I0425 17:44:14.129280 134080978020160 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-true-sv-b6fg5-slice-job-0-0.mt-16-zarr3-true-sv-b6fg5:8482 I0425 17:44:15.019554 134080978020160 max_utils.py:284] Jax distributed system initialized! I0425 17:44:22.115745 134080978020160 max_utils.py:800] System Information: Jax Version: 0.8.1 I0425 17:44:22.115853 134080978020160 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0425 17:44:22.115895 134080978020160 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 I0425 17:44:22.115930 134080978020160 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 17:44:22.738034 134080978020160 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 17:44:22.852403 134080978020160 checkpointing.py:688] Setting up checkpoint logger... I0425 17:44:22.852523 134080978020160 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0425 17:44:22.852568 134080978020160 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0425 17:44:22.852773 134080978020160 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 0x79f1a736cec0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 17:44:25.787026 134080978020160 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0425 17:44:25.787294 134080978020160 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 0x79dd2442e3c0>}, handler_registry=None I0425 17:44:25.787533 134080978020160 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79dd2442e3c0>` 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`. I0425 17:44:25.787582 134080978020160 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79dd244c2fc0>` 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`. I0425 17:44:25.787627 134080978020160 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 0x79dd2442e3c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79dd2442e3c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79dd244c2fc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79dd244c2fc0>}). I0425 17:44:25.788057 134080978020160 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0425 17:44:25.788169 134080978020160 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x79dc847b4f40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0425 17:44:26.922602 134080978020160 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints I0425 17:44:26.932042 134080978020160 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_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79f0645403b0> I0425 17:44:26.932187 134080978020160 checkpointing.py:302] Checkpoint manager created! I0425 17:44:27.890978 134080978020160 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0425 17:44:27.891112 134080978020160 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0425 17:44:29.350372 134080978020160 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0425 17:44:29.350497 134080978020160 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0425 17:44:29.350554 134080978020160 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0425 17:44:29.350598 134080978020160 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0425 17:44:29.350633 134080978020160 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0425 17:44:29.350668 134080978020160 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0425 17:44:29.350704 134080978020160 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 17:44:29.350740 134080978020160 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0425 17:44:29.350774 134080978020160 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 17:44:29.350808 134080978020160 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 17:44:29.350841 134080978020160 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0425 17:44:29.350873 134080978020160 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0425 17:44:29.745197 134080978020160 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 17:44:29.745348 134080978020160 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 17:44:29.750815 134080978020160 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 17:44:29.750910 134080978020160 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 17:44:29.905804 134080978020160 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 17:44:29.905937 134080978020160 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 17:44:30.062795 134080978020160 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 17:44:30.062931 134080978020160 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 17:44:30.227260 134080978020160 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 17:44:30.227402 134080978020160 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 17:44:30.381749 134080978020160 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 17:44:30.381883 134080978020160 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 17:44:30.543730 134080978020160 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 17:44:30.543869 134080978020160 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 17:44:30.708739 134080978020160 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0425 17:44:30.708908 134080978020160 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 17:44:30.732967 134080978020160 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 17:44:30.733063 134080978020160 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 17:44:41.812161 134080978020160 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. I0425 17:44:41.814772 134080978020160 metric_logger.py:289] number parameters: 1.104 billion I0425 17:44:52.985864 134080978020160 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0425 17:44:53.137186 134080978020160 checkpointing.py:798] Waited 0.15131044387817383 seconds for step 0 to finish before starting checkpointing. I0425 17:44:53.139462 134080978020160 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 17:44:53.141149 134080978020160 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0425 17:44:53.143859 134080978020160 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/0. I0425 17:44:53.741578 134080978020160 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 17:44:53.742599 134080978020160 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0425 17:44:53.742661 134080978020160 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 I0425 17:44:53.989064 134080978020160 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.248653s I0425 17:44:53.989264 134080978020160 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.996 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5148756504058838 s) (per-host) I0425 17:44:53.989321 134080978020160 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.514940s (batch_requests_ready=0.259618s, total_serialization_initiated=0.255250s, others=0.000072s) I0425 17:44:53.989418 134080978020160 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.519021s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.519001) I0425 17:44:53.990375 133955047118592 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0425 17:44:53.990562 134080978020160 async_checkpointer.py:561] Finished blocking save. Time taken: 0.849330s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/0. I0425 17:44:54.360155 134080978020160 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0425 17:44:54.360537 133953939805952 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 17:44:54.360702 134080978020160 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777139093.1394436, 'wait_for_prev_duration_secs': 6.270408630371094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777139093.1411896, 'checkpointer_blocking_duration_secs': 0.8494813442230225, 'get_old_steps_start_time': 1777139093.9906921, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1777139093.1376066, 'checkpoint_manager_blocking_duration_secs': 1.2230567932128906} I0425 17:44:54.360814 134080978020160 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0425 17:44:54.360888 134080978020160 max_utils.py:750] Memstats: After params initialized: I0425 17:44:54.360946 134080978020160 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0425 17:44:54.360980 134080978020160 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0425 17:44:54.361009 134080978020160 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0425 17:44:54.361035 134080978020160 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0425 17:44:54.630620 134080978020160 metric_logger.py:185] completed step: 0, seconds: 11.170, TFLOP/s/device: 1.216, Tokens/s/device: 183.348, total_weights: 65536, loss: 10.874 I0425 17:44:54.800564 134080978020160 metric_logger.py:185] completed step: 1, seconds: 1.643, TFLOP/s/device: 8.270, Tokens/s/device: 1246.480, total_weights: 65536, loss: 10.874 I0425 17:44:55.213675 134080978020160 metric_logger.py:185] completed step: 2, seconds: 0.024, TFLOP/s/device: 556.439, Tokens/s/device: 83872.553, total_weights: 65536, loss: 10.560 I0425 17:44:55.361126 134080978020160 metric_logger.py:185] completed step: 3, seconds: 0.404, TFLOP/s/device: 33.624, Tokens/s/device: 5068.153, total_weights: 65536, loss: 9.979 I0425 17:44:55.367053 134080978020160 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0425 17:44:55.656160 134080978020160 checkpointing.py:798] Waited 0.2890799045562744 seconds for step 5 to finish before starting checkpointing. I0425 17:44:55.659245 134080978020160 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 17:44:57.023551 2620 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 17:44:59.491265 133955055511296 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_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/0/items/array_metadatas/process_6 I0425 17:45:31.065225 133955047118592 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 42.021 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.59080648422241 s) (per-host) I0425 17:45:31.065330 133955047118592 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.074841s. I0425 17:45:40.246897 133955047118592 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 46.256392s. I0425 17:45:40.247208 133953939805952 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 17:45:40.247335 133953939805952 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 17:45:40.247386 133953939805952 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0425 17:45:40.249165 133953939805952 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0425 17:45:40.249354 134080978020160 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. W0425 17:45:40.249497 134080978020160 checkpoint_manager.py:1458] Waiting for previous save to complete took 44.590256 seconds. If this number is high, consider checkpointing less frequently. I0425 17:45:40.251342 134080978020160 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 5 I0425 17:45:40.255583 134080978020160 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/5. I0425 17:45:40.935507 134080978020160 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0425 17:45:40.935662 134080978020160 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 I0425 17:45:40.978143 134080978020160 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044738s I0425 17:45:40.978311 134080978020160 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.212 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29596972465515137 s) (per-host) I0425 17:45:40.978366 134080978020160 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.296036s (batch_requests_ready=0.246447s, total_serialization_initiated=0.049515s, others=0.000073s) I0425 17:45:40.978463 134080978020160 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.300766s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.300750) I0425 17:45:40.979398 133953412134656 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0425 17:45:40.979563 134080978020160 async_checkpointer.py:561] Finished blocking save. Time taken: 0.728145s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/5. I0425 17:45:41.011409 134080978020160 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0425 17:45:41.011708 133953939805952 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 17:45:41.011850 134080978020160 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777139095.659216, 'wait_for_prev_duration_secs': 44.59025573730469, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777139140.2513802, 'checkpointer_blocking_duration_secs': 0.7282867431640625, 'get_old_steps_start_time': 1777139140.9796867, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1777139095.656508, 'checkpoint_manager_blocking_duration_secs': 45.355308294296265} I0425 17:45:41.011967 134080978020160 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0425 17:45:41.012770 134080978020160 metric_logger.py:185] completed step: 4, seconds: 0.162, TFLOP/s/device: 83.615, Tokens/s/device: 12603.387, total_weights: 65536, loss: 9.457 I0425 17:45:41.020157 134080978020160 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.367, Tokens/s/device: 13922.502, total_weights: 65536, loss: 8.955 I0425 17:46:00.851445 134080978020160 metric_logger.py:185] completed step: 6, seconds: 45.651, TFLOP/s/device: 0.298, Tokens/s/device: 44.862, total_weights: 65536, loss: 8.465 I0425 17:46:00.998766 134080978020160 metric_logger.py:185] completed step: 7, seconds: 19.686, TFLOP/s/device: 0.690, Tokens/s/device: 104.033, total_weights: 65536, loss: 7.998 I0425 17:46:01.146388 134080978020160 metric_logger.py:185] completed step: 8, seconds: 0.154, TFLOP/s/device: 88.488, Tokens/s/device: 13337.936, total_weights: 65536, loss: 7.567 I0425 17:46:01.294770 134080978020160 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0425 17:46:01.295381 134080978020160 checkpointing.py:798] Waited 0.0006213188171386719 seconds for step 9 to finish before starting checkpointing. I0425 17:46:01.297322 134080978020160 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 17:46:05.667263 133955055511296 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_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/5/items/array_metadatas/process_6 I0425 17:46:40.393263 133953412134656 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 26.454 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 59.71088361740112 s) (per-host) I0425 17:46:40.393390 133953412134656 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 59.413879s. I0425 17:46:50.241605 133953412134656 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 69.262079s. I0425 17:46:50.241893 133953939805952 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 17:46:50.242035 133953939805952 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 17:46:50.242086 133953939805952 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0425 17:46:50.243429 133953939805952 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0425 17:46:50.243595 134080978020160 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. W0425 17:46:50.243726 134080978020160 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.946416 seconds. If this number is high, consider checkpointing less frequently. I0425 17:46:50.245331 134080978020160 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0425 17:46:50.248857 134080978020160 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/9. I0425 17:46:50.834335 134080978020160 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0425 17:46:50.834434 134080978020160 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 I0425 17:46:50.876928 134080978020160 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044624s I0425 17:46:50.877108 134080978020160 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.763 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.32387328147888184 s) (per-host) I0425 17:46:50.877161 134080978020160 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.323950s (batch_requests_ready=0.274754s, total_serialization_initiated=0.049114s, others=0.000082s) I0425 17:46:50.877261 134080978020160 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.327854s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.327842) I0425 17:46:50.878132 133955047118592 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0425 17:46:50.878295 134080978020160 async_checkpointer.py:561] Finished blocking save. Time taken: 0.632895s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/9. I0425 17:46:50.880184 134080978020160 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0425 17:46:50.880396 133953939805952 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 17:46:50.880507 134080978020160 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777139161.2972934, 'wait_for_prev_duration_secs': 48.94641590118408, 'time_between_consecutive_saves_sec': 21.048088788986206, 'checkpointer_blocking_start_time': 1777139210.24537, 'checkpointer_blocking_duration_secs': 0.6330344676971436, 'get_old_steps_start_time': 1777139210.8784173, 'get_old_steps_duration_secs': 1.8596649169921875e-05, 'checkpoint_manager_blocking_start_time': 1777139161.2956235, 'checkpoint_manager_blocking_duration_secs': 49.58485150337219} I0425 17:46:50.880623 134080978020160 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0425 17:46:50.880665 134080978020160 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 17:46:56.075398 133955055511296 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_20260425_150412/nnx_xpk_feat_nnx_set_defaults_true_20260425_150412_16_zarr3_true/checkpoints/9/items/array_metadatas/process_6 I0425 17:47:32.402611 133955047118592 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 37.745 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.849353075027466 s) (per-host) I0425 17:47:32.402732 133955047118592 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.524488s. I0425 17:47:40.259715 133955047118592 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 49.381456s. I0425 17:47:40.259941 133953939805952 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 17:47:40.260004 133953939805952 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 17:47:40.260047 133953939805952 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0425 17:47:40.261642 133953939805952 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0425 17:47:40.261889 134080978020160 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. I0425 17:47:40.262086 134080978020160 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 17:47:40.262820 134080978020160 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.737, Tokens/s/device: 13978.377, 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: Sat Apr 25 17:47:51 UTC 2026 EXIT_CODE=0