XPK Start: Thu Apr 16 23:08:22 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-16 23:08:46.319463: 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) I0416 23:08:46.499360 133919886260032 max_utils.py:273] Attempting to initialize the jax distributed system... I0416 23:08:55.539159 133919886260032 distributed.py:149] Starting JAX distributed service on [::]:8482 I0416 23:08:55.541445 133919886260032 distributed.py:172] Connecting to JAX distributed service on mt-13-scan-layers-true-5tl4f-slice-job-0-0.mt-13-scan-layers-true-5tl4f:8482 I0416 23:08:57.220618 133919886260032 max_utils.py:284] Jax distributed system initialized! I0416 23:09:03.410392 133919886260032 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 23:09:03.410492 133919886260032 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 23:09:03.410532 133919886260032 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0416 23:09:03.410562 133919886260032 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 23:09:04.074721 133919886260032 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 23:09:04.191393 133919886260032 checkpointing.py:688] Setting up checkpoint logger... I0416 23:09:04.191511 133919886260032 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 23:09:04.191557 133919886260032 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0416 23:09:04.191755 133919886260032 base_pytree_checkpoint_handler.py:441] 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 0x79cbef84c770>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 23:09:07.453717 133919886260032 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 23:09:07.454022 133919886260032 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b7f47b7a40>}, handler_registry=None I0416 23:09:07.454268 133919886260032 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b7f47b7a40>` 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`. I0416 23:09:07.454317 133919886260032 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b7f4766a20>` 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`. I0416 23:09:07.454352 133919886260032 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 0x79b7f47b7a40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b7f47b7a40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b7f4766a20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b7f4766a20>}). I0416 23:09:07.454666 133919886260032 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0416 23:09:07.454736 133919886260032 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x79b7f46b2020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0416 23:09:08.703562 133919886260032 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints I0416 23:09:09.542736 133919886260032 checkpoint_manager.py:929] [process=0][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_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79b8146ca660> I0416 23:09:09.542994 133919886260032 checkpointing.py:302] Checkpoint manager created! I0416 23:09:10.485729 133919886260032 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 23:09:10.485857 133919886260032 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0416 23:09:11.882956 133919886260032 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 23:09:11.883070 133919886260032 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 23:09:11.883121 133919886260032 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 23:09:11.883162 133919886260032 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 23:09:11.883199 133919886260032 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 23:09:11.883233 133919886260032 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 23:09:11.883269 133919886260032 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 23:09:11.883304 133919886260032 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 23:09:11.883338 133919886260032 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 23:09:11.883372 133919886260032 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 23:09:11.883404 133919886260032 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 23:09:11.883437 133919886260032 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 23:09:12.296987 133919886260032 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 23:09:12.297076 133919886260032 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 23:09:12.302552 133919886260032 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 23:09:12.302608 133919886260032 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 23:09:12.319032 133919886260032 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 23:09:12.319094 133919886260032 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 23:09:12.335217 133919886260032 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 23:09:12.335274 133919886260032 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0416 23:09:12.359745 133919886260032 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 23:09:12.359809 133919886260032 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 23:09:12.376356 133919886260032 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 23:09:12.376420 133919886260032 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 23:09:12.392406 133919886260032 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 23:09:12.392471 133919886260032 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 23:09:12.423016 133919886260032 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 23:09:12.423091 133919886260032 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0416 23:09:12.442993 133919886260032 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 23:09:12.443054 133919886260032 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0416 23:09:23.691012 133919886260032 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. I0416 23:09:23.700123 133919886260032 metric_logger.py:289] number parameters: 1.104 billion I0416 23:09:35.229783 133919886260032 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 23:09:35.382677 133919886260032 checkpointing.py:798] Waited 0.1528797149658203 seconds for step 0 to finish before starting checkpointing. I0416 23:09:35.384961 133919886260032 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 23:09:35.386480 133919886260032 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0416 23:09:35.387817 133919886260032 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0. I0416 23:09:36.100249 133919886260032 signaling_client.py:364] Using JaxDistributedSignalingClient I0416 23:09:36.120299 133919886260032 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 23:09:36.120369 133919886260032 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 I0416 23:09:36.352770 133791735330560 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0 I0416 23:09:36.395553 133919886260032 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.276678s I0416 23:09:36.396337 133919886260032 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.357 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2879505157470703 s) (per-host) I0416 23:09:36.396398 133919886260032 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288028s (batch_requests_ready=0.003199s, total_serialization_initiated=0.284135s, others=0.000694s) I0416 23:09:36.396474 133919886260032 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292173s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.292155) I0416 23:09:36.397289 133919886260032 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0. I0416 23:09:36.397587 133791743723264 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 23:29:36.397555 I0416 23:09:36.401329 133919886260032 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 23:09:36.401567 133789095290624 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 23:09:36.401684 133919886260032 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776380975.3849423, 'wait_for_prev_duration_secs': 5.91278076171875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776380975.3865173, 'checkpointer_blocking_duration_secs': 1.0112051963806152, 'get_old_steps_start_time': 1776380976.3977454, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776380975.3830876, 'checkpoint_manager_blocking_duration_secs': 1.0185585021972656} I0416 23:09:36.401795 133919886260032 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 23:09:36.401883 133919886260032 max_utils.py:750] Memstats: After params initialized: I0416 23:09:36.401942 133919886260032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0416 23:09:36.401976 133919886260032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0416 23:09:36.402001 133919886260032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0416 23:09:36.402025 133919886260032 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0416 23:09:36.740957 133919886260032 metric_logger.py:185] completed step: 0, seconds: 11.204, TFLOP/s/device: 1.213, Tokens/s/device: 182.792, total_weights: 65536, loss: 10.874 I0416 23:09:36.743865 133919886260032 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/tensorboard/' I0416 23:09:37.588508 133919886260032 metric_logger.py:185] completed step: 1, seconds: 1.508, TFLOP/s/device: 9.012, Tokens/s/device: 1358.393, total_weights: 65536, loss: 10.874 I0416 23:09:37.730051 133919886260032 metric_logger.py:185] completed step: 2, seconds: 0.848, TFLOP/s/device: 16.014, Tokens/s/device: 2413.856, total_weights: 65536, loss: 10.262 I0416 23:09:37.877453 133919886260032 metric_logger.py:185] completed step: 3, seconds: 0.013, TFLOP/s/device: 1079.029, Tokens/s/device: 162642.948, total_weights: 65536, loss: 9.732 I0416 23:09:38.124228 133791242774272 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/items I0416 23:09:38.263555 133787463706368 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776380977869533825, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776380977869533825, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0416 23:09:39.927499 2804 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0416 23:09:42.130244 133790684931840 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/items/array_metadatas/process_0 I0416 23:10:01.331469 133919886260032 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 95.193, Tokens/s/device: 14348.469, total_weights: 65536, loss: 9.274 I0416 23:10:01.341135 133919886260032 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.583, Tokens/s/device: 13955.136, total_weights: 65536, loss: 8.888 I0416 23:10:01.485888 133919886260032 metric_logger.py:185] completed step: 6, seconds: 23.452, TFLOP/s/device: 0.579, Tokens/s/device: 87.327, total_weights: 65536, loss: 8.589 I0416 23:10:01.633210 133919886260032 metric_logger.py:185] completed step: 7, seconds: 0.007, TFLOP/s/device: 1866.364, Tokens/s/device: 281318.681, total_weights: 65536, loss: 8.380 I0416 23:10:01.780755 133919886260032 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 92.813, Tokens/s/device: 13989.836, total_weights: 65536, loss: 8.251 I0416 23:10:01.929180 133919886260032 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 23:10:01.929877 133919886260032 checkpointing.py:798] Waited 0.0007083415985107422 seconds for step 9 to finish before starting checkpointing. I0416 23:10:01.931954 133919886260032 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 23:10:16.420608 133789103683328 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.502259s (commit=35.336993s, array_metadata_write=1.165266s) I0416 23:10:16.421659 133791743723264 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.183 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.31322979927063 s) (per-host) I0416 23:10:16.421781 133791743723264 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.024089s. I0416 23:10:17.960495 133791743723264 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776380977869533825, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0416 23:10:18.707803 133791743723264 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000423s. I0416 23:10:18.708136 133787463706368 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776380977869533825, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0416 23:10:19.317574 133791743723264 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 23:10:20.238587 133791743723264 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.152183s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/items I0416 23:10:20.240239 133791743723264 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0/items I0416 23:10:20.798295 133791743723264 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0 I0416 23:10:22.212583 133791743723264 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0`. I0416 23:10:22.213262 133791743723264 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.83 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/0 I0416 23:10:22.214826 133791743723264 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.817132s. I0416 23:10:22.215038 133789095290624 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 23:10:22.215154 133789095290624 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 23:10:22.215217 133789095290624 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 23:10:22.217816 133789095290624 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 23:10:22.217968 133919886260032 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0416 23:10:22.218056 133919886260032 checkpoint_manager.py:1452] Waiting for previous save to complete took 20.286108 seconds. If this number is high, consider checkpointing less frequently. I0416 23:10:22.220059 133919886260032 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0416 23:10:22.222187 133919886260032 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9. I0416 23:10:22.940729 133919886260032 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 23:10:22.940820 133919886260032 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 I0416 23:10:22.967685 133919886260032 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028462s I0416 23:10:22.968522 133919886260032 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.984 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03763866424560547 s) (per-host) I0416 23:10:22.968590 133919886260032 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037724s (batch_requests_ready=0.002807s, total_serialization_initiated=0.034127s, others=0.000789s) I0416 23:10:22.968677 133919886260032 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041612s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.041595) I0416 23:10:22.969376 133919886260032 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9. I0416 23:10:22.969668 133790689117952 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 23:30:22.969633 I0416 23:10:22.976219 133919886260032 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 23:10:22.976462 133789095290624 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 23:10:22.976588 133919886260032 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776381001.9319212, 'wait_for_prev_duration_secs': 20.286107778549194, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776381022.2200992, 'checkpointer_blocking_duration_secs': 0.7497096061706543, 'get_old_steps_start_time': 1776381022.9698474, 'get_old_steps_duration_secs': 3.647804260253906e-05, 'checkpoint_manager_blocking_start_time': 1776381001.9300997, 'checkpoint_manager_blocking_duration_secs': 21.04645586013794} I0416 23:10:22.976692 133919886260032 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 23:10:22.976737 133919886260032 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 23:10:23.170894 133791743723264 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9 I0416 23:10:24.898932 133790705903360 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9/items I0416 23:10:27.760816 133790714296064 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9/items/array_metadatas/process_0 I0416 23:11:07.406341 133790697510656 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.390488s (commit=40.195739s, array_metadata_write=1.194750s) I0416 23:11:07.407478 133790689117952 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.515 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.47656559944153 s) (per-host) I0416 23:11:07.407602 133790689117952 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.437823s. I0416 23:11:09.075700 133790689117952 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000407s. I0416 23:11:09.474823 133790689117952 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 23:11:10.387973 133790689117952 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.155645s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9/items I0416 23:11:10.389670 133790689117952 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9/items I0416 23:11:10.984154 133790689117952 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9 I0416 23:11:12.365542 133790689117952 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9`. I0416 23:11:12.366463 133790689117952 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.15 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_13_scan_layers_true/checkpoints/9 I0416 23:11:12.367935 133790689117952 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.398154s. I0416 23:11:12.368055 133789095290624 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 23:11:12.368120 133789095290624 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 23:11:12.368195 133789095290624 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 23:11:12.370276 133789095290624 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 23:11:12.370453 133919886260032 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0416 23:11:12.370626 133919886260032 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 23:11:12.371304 133919886260032 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.436, Tokens/s/device: 13932.921, 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 16 23:11:23 UTC 2026 EXIT_CODE=0