2026-04-16 19:27:07.718902: 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 19:27:07.836747 129362090228864 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 19:28:03.388469 129362090228864 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 19:28:03.388574 129362090228864 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 19:28:03.388620 129362090228864 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0416 19:28:03.388647 129362090228864 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 19:28:03.388673 129362090228864 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 19:28:03.388771 129362090228864 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 19:28:03.484545 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:03.888376 129362090228864 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:28:03.888542 129362090228864 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:28:03.888782 129362090228864 checkpointing.py:688] Setting up checkpoint logger... I0416 19:28:03.888826 129362090228864 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 19:28:03.888865 129362090228864 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 19:28:03.889455 129362090228864 base_pytree_checkpoint_handler.py:411] 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 0x75a6de125a30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 19:28:05.554565 129362090228864 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 19:28:05.554743 129362090228864 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75a1e0865bb0>}, handler_registry=None I0416 19:28:05.555006 129362090228864 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75a1e0865bb0>` 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 19:28:05.555060 129362090228864 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75a062d3c800>` 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 19:28:05.555094 129362090228864 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 0x75a1e0865bb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75a1e0865bb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75a062d3c800>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75a062d3c800>}). I0416 19:28:05.555512 129362090228864 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 19:28:05.555562 129362090228864 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x75a062d31ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 19:28:05.964763 129255956547136 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367542859637477, 'commit_timestamp_nsecs': 1776367598052821853, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0416 19:28:06.046772 129255956547136 checkpoint_manager.py:1847] Missing metrics for step 9 I0416 19:28:06.113654 129362090228864 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints I0416 19:28:06.114131 129362090228864 checkpoint_manager.py:921] [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_hns=False, 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), root_directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x75a132d56000> I0416 19:28:06.114219 129362090228864 checkpointing.py:302] Checkpoint manager created! W0416 19:28:06.130723 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:06.317822 129362090228864 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 19:28:06.317914 129362090228864 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:28:06.412448 129362090228864 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:28:06.412527 129362090228864 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:28:06.426821 129362090228864 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:28:06.426876 129362090228864 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:28:06.448061 129362090228864 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 19:28:06.448118 129362090228864 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:28:06.462401 129362090228864 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 19:28:06.462451 129362090228864 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:28:06.476644 129362090228864 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 19:28:06.476690 129362090228864 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:28:06.498175 129362090228864 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 19:28:06.498228 129362090228864 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:28:06.516073 129362090228864 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 19:28:06.516126 129362090228864 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 19:28:06.702986 129362090228864 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 19:28:06.703087 129362090228864 checkpointing.py:582] restoring from this run's directory step 9 I0416 19:28:06.872772 129362090228864 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9. I0416 19:28:07.408591 4038140 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 19:28:07.408657 4038140 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 19:28:13.492129 129362090228864 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.969 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 19:28:13.492457 129362090228864 checkpointer.py:318] Finished restoring checkpoint in 6.71 seconds from gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9. I0416 19:28:13.492561 129362090228864 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776367686.7873504, 'checkpointer_duration_secs': 6.705162286758423, 'checkpoint_manager_start_time': 1776367686.7033129, 'checkpoint_manager_duration_secs': 6.789202451705933} [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0416 19:28:14.149197 129362090228864 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 19:28:14.149312 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 19:28:14.149354 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 19:28:14.149398 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0416 19:28:14.149438 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 19:28:14.149465 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 19:28:14.149504 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0416 19:28:14.149549 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0416 19:28:14.149579 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0416 19:28:14.149605 129362090228864 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0416 19:28:14.149641 129362090228864 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 19:28:14.149678 129362090228864 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 19:28:14.598634 129362090228864 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:28:14.598714 129362090228864 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 19:28:14.612671 129362090228864 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:28:14.612719 129362090228864 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 19:28:15.283733 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:15.348847 129362090228864 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0416 19:28:15.349289 129362090228864 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 19:28:15.349435 129362090228864 metric_logger.py:289] number parameters: 1.104 billion W0416 19:28:16.200433 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:16.271375 129362090228864 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0416 19:28:16.388652 129362090228864 checkpointing.py:798] Waited 0.1172642707824707 seconds for step 10 to finish before starting checkpointing. I0416 19:28:16.389079 129362090228864 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:28:16.389242 129362090228864 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0416 19:28:16.389529 129362090228864 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10. I0416 19:28:16.466563 129362090228864 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 19:28:16.554653 129254400460352 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10 I0416 19:28:16.564022 129362090228864 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:28:16.564128 129362090228864 replica_slices.py:410] 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 W0416 19:28:16.785739 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.801125 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.820363 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.826293 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.830250 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.834139 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.837994 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:16.841810 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:17.276097 129254389974592 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/items I0416 19:28:17.425307 129254358517312 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367697190449522, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776367697190449522, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0416 19:28:25.832906 129362090228864 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.269303s I0416 19:28:25.837282 129362090228864 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.317 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0416 19:28:25.837343 129362090228864 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.369874s (batch_requests_ready=0.093393s, total_serialization_initiated=9.272219s, others=0.004261s) I0416 19:28:25.837411 129362090228864 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.370403s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=9.370384) I0416 19:28:25.838816 129254379488832 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:28:25.838897 129362090228864 async_checkpointer.py:561] Finished blocking save. Time taken: 9.449615s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10. I0416 19:28:25.839092 129362090228864 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:28:25.839256 129254410946112 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:28:25.839344 129362090228864 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367696.389033, 'wait_for_prev_duration_secs': 7.343292236328125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367696.3892636, 'checkpointer_blocking_duration_secs': 9.449737310409546, 'get_old_steps_start_time': 1776367705.8390167, 'get_old_steps_duration_secs': 3.933906555175781e-05, 'checkpoint_manager_blocking_start_time': 1776367696.38895, 'checkpoint_manager_blocking_duration_secs': 9.450367450714111} I0416 19:28:25.839456 129362090228864 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0416 19:28:25.839494 129362090228864 max_utils.py:750] Memstats: After params initialized: I0416 19:28:25.839536 129362090228864 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 19:28:25.839560 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 19:28:25.839586 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 19:28:25.839604 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 19:28:25.839622 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 19:28:25.839638 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 19:28:25.839654 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 19:28:25.839669 129362090228864 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 19:28:25.845648 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:25.851556 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:28:25.855731 4036469 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:28:26.043204 129362090228864 metric_logger.py:185] completed step: 10, seconds: 0.921, TFLOP/s/device: 14.745, Tokens/s/device: 2222.514, total_weights: 16384, loss: 3.602 I0416 19:28:26.159582 129362090228864 metric_logger.py:185] completed step: 11, seconds: 9.770, TFLOP/s/device: 1.391, Tokens/s/device: 209.614, total_weights: 16384, loss: 3.101 I0416 19:28:26.278058 129362090228864 metric_logger.py:185] completed step: 12, seconds: 0.005, TFLOP/s/device: 2739.893, Tokens/s/device: 412986.489, total_weights: 16384, loss: 2.684 I0416 19:28:26.396563 129362090228864 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.750, Tokens/s/device: 17447.160, total_weights: 16384, loss: 2.348 I0416 19:28:26.401167 129362090228864 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0416 19:28:26.632955 129362090228864 checkpointing.py:798] Waited 0.23175883293151855 seconds for step 15 to finish before starting checkpointing. I0416 19:28:26.633355 129362090228864 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:28:26.648519 129254369003072 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0 I0416 19:29:06.040368 129254389974592 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.202362s (commit=39.774905s, array_metadata_write=0.427457s) I0416 19:29:06.041326 129254379488832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 254.908 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 49 seconds) (per-host) I0416 19:29:06.041383 129254379488832 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.202443s. I0416 19:29:06.448874 129254379488832 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 19:29:06.654664 129254358517312 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367697190449522, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0416 19:29:06.874740 129254379488832 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:29:06.875304 129254379488832 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570701s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/items I0416 19:29:06.875984 129254379488832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10/items I0416 19:29:07.109964 129254379488832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10 I0416 19:29:07.795686 129254379488832 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10`. I0416 19:29:07.796304 129254379488832 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 51.407029s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/10 I0416 19:29:07.796388 129254379488832 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.957448s. I0416 19:29:07.796574 129254410946112 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:29:07.796686 129254410946112 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:29:07.796753 129254410946112 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:29:07.796806 129254410946112 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0416 19:29:07.796966 129362090228864 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0416 19:29:07.797093 129362090228864 checkpoint_manager.py:1441] Waiting for previous save to complete took 41.163738 seconds. If this number is high, consider checkpointing less frequently. I0416 19:29:07.797693 129362090228864 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0416 19:29:07.797971 129362090228864 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15. I0416 19:29:07.960375 129254410946112 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15 I0416 19:29:07.969758 129362090228864 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:29:07.969855 129362090228864 replica_slices.py:410] 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 19:29:08.701555 129254389974592 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15/items I0416 19:29:13.245312 129362090228864 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.275928s I0416 19:29:13.245836 129362090228864 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.302 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0416 19:29:13.245895 129362090228864 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.360672s (batch_requests_ready=0.080961s, total_serialization_initiated=5.279296s, others=0.000414s) I0416 19:29:13.245957 129362090228864 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.361185s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=5.361172) I0416 19:29:13.250870 129255904118336 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:29:13.250955 129362090228864 async_checkpointer.py:561] Finished blocking save. Time taken: 5.453214s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15. I0416 19:29:13.251155 129362090228864 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:29:13.251328 129254379488832 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:29:13.251436 129362090228864 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367706.6333325, 'wait_for_prev_duration_secs': 41.16373801231384, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367747.7977204, 'checkpointer_blocking_duration_secs': 5.4533562660217285, 'get_old_steps_start_time': 1776367753.2510927, 'get_old_steps_duration_secs': 2.1219253540039062e-05, 'checkpoint_manager_blocking_start_time': 1776367706.6332698, 'checkpoint_manager_blocking_duration_secs': 46.618141174316406} I0416 19:29:13.251546 129362090228864 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0416 19:29:13.252023 129362090228864 metric_logger.py:185] completed step: 14, seconds: 0.118, TFLOP/s/device: 114.786, Tokens/s/device: 17301.827, total_weights: 16384, loss: 2.083 I0416 19:29:13.259393 129362090228864 metric_logger.py:185] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.551, Tokens/s/device: 17266.381, total_weights: 16384, loss: 1.880 I0416 19:29:13.375683 129362090228864 metric_logger.py:185] completed step: 16, seconds: 46.857, TFLOP/s/device: 0.290, Tokens/s/device: 43.708, total_weights: 16384, loss: 1.726 I0416 19:29:13.494375 129362090228864 metric_logger.py:185] completed step: 17, seconds: 0.005, TFLOP/s/device: 2659.970, Tokens/s/device: 400939.702, total_weights: 16384, loss: 1.610 I0416 19:29:13.612954 129362090228864 metric_logger.py:185] completed step: 18, seconds: 0.118, TFLOP/s/device: 115.355, Tokens/s/device: 17387.613, total_weights: 16384, loss: 1.521 I0416 19:29:13.731637 129362090228864 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0416 19:29:13.732526 129362090228864 checkpointing.py:798] Waited 0.0009105205535888672 seconds for step 19 to finish before starting checkpointing. I0416 19:29:13.732763 129362090228864 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:29:13.976379 129255946061376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0 I0416 19:29:55.921617 129255935575616 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.671326s (commit=42.218796s, array_metadata_write=0.452530s) I0416 19:29:55.922724 129255904118336 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 263.061 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0416 19:29:55.922782 129255904118336 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.671782s. I0416 19:29:56.321220 129255904118336 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 19:29:56.748867 129255904118336 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:29:56.749032 129255904118336 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573475s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15/items I0416 19:29:56.749700 129255904118336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15/items I0416 19:29:56.979231 129255904118336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15 I0416 19:29:57.659219 129255904118336 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15`. I0416 19:29:57.659831 129255904118336 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.862099s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/15 I0416 19:29:57.659906 129255904118336 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 44.408907s. I0416 19:29:57.660081 129254379488832 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:29:57.660182 129254379488832 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:29:57.660252 129254379488832 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:29:57.660293 129254379488832 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0416 19:29:57.660432 129362090228864 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0416 19:29:57.660543 129362090228864 checkpoint_manager.py:1441] Waiting for previous save to complete took 43.927786 seconds. If this number is high, consider checkpointing less frequently. I0416 19:29:57.660772 129362090228864 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0416 19:29:57.661073 129362090228864 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19. I0416 19:29:57.826677 129254379488832 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19 I0416 19:29:57.840770 129362090228864 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:29:57.840869 129362090228864 replica_slices.py:410] 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 19:29:58.532682 129254410946112 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19/items I0416 19:30:09.463466 129362090228864 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.623087s I0416 19:30:09.467759 129362090228864 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.052 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 19:30:09.467816 129362090228864 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.725389s (batch_requests_ready=0.094445s, total_serialization_initiated=11.626789s, others=0.004155s) I0416 19:30:09.467908 129362090228864 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.725922s (all_items=0.000014s, per_item={'items': '0.00001359'}, temp_paths=11.725908) I0416 19:30:09.469550 129255904118336 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:30:09.469658 129362090228864 async_checkpointer.py:561] Finished blocking save. Time taken: 11.808842s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19. I0416 19:30:09.469912 129362090228864 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:30:09.470084 129255893632576 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:30:09.470190 129362090228864 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367753.73274, 'wait_for_prev_duration_secs': 43.927786111831665, 'time_between_consecutive_saves_sec': 5.935919761657715, 'checkpointer_blocking_start_time': 1776367797.6607933, 'checkpointer_blocking_duration_secs': 11.809040069580078, 'get_old_steps_start_time': 1776367809.469852, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776367753.7327085, 'checkpoint_manager_blocking_duration_secs': 55.737454652786255} I0416 19:30:09.470334 129362090228864 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0416 19:30:09.470364 129362090228864 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:30:10.210686 129255946061376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0 I0416 19:30:46.682341 129255935575616 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.213703s (commit=36.805646s, array_metadata_write=0.408057s) I0416 19:30:46.683389 129255904118336 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 258.205 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0416 19:30:46.683497 129255904118336 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.213748s. I0416 19:30:47.087337 129255904118336 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 19:30:47.483363 129255904118336 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:30:47.483957 129255904118336 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.536373s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19/items I0416 19:30:47.484589 129255904118336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19/items I0416 19:30:47.730573 129255904118336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19 I0416 19:30:48.428013 129255904118336 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19`. I0416 19:30:48.428638 129255904118336 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.767836s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/19 I0416 19:30:48.428716 129255904118336 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.958969s. I0416 19:30:48.428887 129255893632576 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:30:48.428992 129255893632576 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:30:48.429063 129255893632576 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:30:48.429110 129255893632576 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0416 19:30:48.429258 129362090228864 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0416 19:30:48.429388 129362090228864 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:30:48.429967 129362090228864 metric_logger.py:185] completed step: 19, seconds: 0.118, TFLOP/s/device: 114.937, Tokens/s/device: 17324.513, total_weights: 16384, loss: 1.449 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops