2026-04-16 20:49:06.318813: 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 20:49:06.434999 138808848891008 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. ~/maxtext_venv/lib/python3.12/site-packages/jax/_src/xla_bridge.py:219: UserWarning: TPU backend initialization is taking more than 60.0 seconds. Did you run your code on all TPU hosts? See https://docs.jax.dev/en/latest/multi_process.html for more information. warnings.warn( I0416 20:50:11.714491 138808848891008 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 20:50:11.714615 138808848891008 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 20:50:11.714653 138808848891008 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 20:50:11.714678 138808848891008 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 20:50:11.714700 138808848891008 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 20:50:11.714777 138808848891008 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 20:50:11.811733 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:12.222886 138808848891008 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 20:50:12.323010 138808848891008 checkpointing.py:688] Setting up checkpoint logger... I0416 20:50:12.323133 138808848891008 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 20:50:12.323202 138808848891008 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 20:50:12.323426 138808848891008 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 0x7e3e5ddfcf80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 20:50:14.704571 138808848891008 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 20:50:14.704844 138808848891008 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 0x7e37e05bcec0>}, handler_registry=None I0416 20:50:14.705209 138808848891008 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e37e05bcec0>` 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 20:50:14.705253 138808848891008 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e37e05742f0>` 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 20:50:14.705283 138808848891008 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 0x7e37e05bcec0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e37e05bcec0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e37e05742f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e37e05742f0>}). I0416 20:50:14.705627 138808848891008 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 20:50:14.705686 138808848891008 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7e37e1599800> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 20:50:14.824754 138808848891008 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints I0416 20:50:14.825006 138808848891008 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/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e37e0531820> I0416 20:50:14.825118 138808848891008 checkpointing.py:302] Checkpoint manager created! W0416 20:50:14.842325 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:15.060395 138808848891008 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 20:50:15.060515 138808848891008 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 20:50:15.172692 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 20:50:15.261444 138808848891008 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 20:50:15.261555 138808848891008 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:50:15.261596 138808848891008 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:50:15.261630 138808848891008 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 20:50:15.261658 138808848891008 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:50:15.261684 138808848891008 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:50:15.261710 138808848891008 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:50:15.261737 138808848891008 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 20:50:15.261763 138808848891008 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:50:15.261787 138808848891008 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:50:15.261811 138808848891008 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 20:50:15.261833 138808848891008 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 20:50:15.373579 138808848891008 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:50:15.373671 138808848891008 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:50:15.379165 138808848891008 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:50:15.379213 138808848891008 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:50:15.393845 138808848891008 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:50:15.393897 138808848891008 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:50:15.407739 138808848891008 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:50:15.407793 138808848891008 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:50:15.429279 138808848891008 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 20:50:15.429341 138808848891008 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:50:15.443350 138808848891008 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 20:50:15.443403 138808848891008 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:50:15.457329 138808848891008 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 20:50:15.457381 138808848891008 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:50:15.484869 138808848891008 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 20:50:15.484930 138808848891008 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:50:15.503454 138808848891008 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 20:50:15.503522 138808848891008 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). W0416 20:50:16.235406 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:16.357766 138808848891008 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 20:50:16.358426 138808848891008 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 20:50:16.359780 138808848891008 metric_logger.py:289] number parameters: 1.104 billion W0416 20:50:16.999074 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:17.097162 138808848891008 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 20:50:17.213288 138808848891008 checkpointing.py:798] Waited 0.1161041259765625 seconds for step 0 to finish before starting checkpointing. I0416 20:50:17.213773 138808848891008 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:50:17.213929 138808848891008 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 20:50:17.214379 138808848891008 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0. I0416 20:50:17.305329 138808848891008 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 20:50:17.389122 138701413811776 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0 I0416 20:50:17.400082 138808848891008 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:50:17.400185 138808848891008 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 20:50:18.028651 138701403326016 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/items W0416 20:50:18.088588 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.098701 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.115958 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.125973 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.136594 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.142810 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.147686 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:18.151951 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:18.183863 138701355091520 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372617938124812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776372617938124812, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:50:18.573400 4132849 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 20:50:18.573479 4132849 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 20:50:20.032747 138808848891008 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.633790s I0416 20:50:20.039216 138808848891008 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.517 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0416 20:50:20.039285 138808848891008 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.731928s (batch_requests_ready=0.083897s, total_serialization_initiated=2.641695s, others=0.006336s) I0416 20:50:20.039368 138808848891008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.732919s (all_items=0.000022s, per_item={'items': '0.00002193'}, temp_paths=2.732897) I0416 20:50:20.040998 138701424297536 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:50:20.041125 138808848891008 async_checkpointer.py:561] Finished blocking save. Time taken: 2.827157s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0. I0416 20:50:20.041314 138808848891008 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:50:20.041480 138701271205440 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:50:20.041577 138808848891008 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_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': 1776372617.21376, 'wait_for_prev_duration_secs': 4.029273986816406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372617.2139494, 'checkpointer_blocking_duration_secs': 2.8272883892059326, 'get_old_steps_start_time': 1776372620.0412564, 'get_old_steps_duration_secs': 2.1696090698242188e-05, 'checkpoint_manager_blocking_start_time': 1776372617.2136698, 'checkpoint_manager_blocking_duration_secs': 2.8278815746307373} I0416 20:50:20.041693 138808848891008 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 20:50:20.041752 138808848891008 max_utils.py:750] Memstats: After params initialized: I0416 20:50:20.041802 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 20:50:20.041826 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 20:50:20.041848 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 20:50:20.041870 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 20:50:20.041887 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 20:50:20.041904 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 20:50:20.041922 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 20:50:20.041938 138808848891008 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 20:50:20.049023 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:20.055151 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 20:50:20.059968 4129610 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:50:20.264019 138808848891008 metric_logger.py:185] completed step: 0, seconds: 0.736, TFLOP/s/device: 18.458, Tokens/s/device: 2782.193, total_weights: 16384, loss: 10.881 I0416 20:50:20.264845 138808848891008 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/tensorboard/' I0416 20:50:20.378598 138808848891008 metric_logger.py:185] completed step: 1, seconds: 3.165, TFLOP/s/device: 4.294, Tokens/s/device: 647.166, total_weights: 16384, loss: 10.881 I0416 20:50:20.497188 138808848891008 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 1336.001, Tokens/s/device: 201376.598, total_weights: 16384, loss: 10.282 I0416 20:50:20.615659 138808848891008 metric_logger.py:185] completed step: 3, seconds: 0.112, TFLOP/s/device: 120.785, Tokens/s/device: 18206.063, total_weights: 16384, loss: 9.282 I0416 20:50:20.621628 138808848891008 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0416 20:50:20.762896 138701382354496 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/items/array_metadatas/process_0 I0416 20:50:20.852365 138808848891008 checkpointing.py:798] Waited 0.23070168495178223 seconds for step 5 to finish before starting checkpointing. I0416 20:50:20.852805 138808848891008 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:50:51.253024 138701292176960 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.212921s (commit=30.693074s, array_metadata_write=0.519847s) I0416 20:50:51.254383 138701424297536 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 372.251 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0416 20:50:51.254525 138701424297536 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.213348s. I0416 20:50:51.477036 138701424297536 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372617938124812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:50:51.660553 138701424297536 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 20:50:51.852955 138701355091520 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776372617938124812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0416 20:50:52.095356 138701424297536 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:50:52.095566 138701424297536 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.576614s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/items I0416 20:50:52.096339 138701424297536 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0/items I0416 20:50:52.336529 138701424297536 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0 I0416 20:50:52.999131 138701424297536 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0`. I0416 20:50:52.999879 138701424297536 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.785911s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/0 I0416 20:50:52.999953 138701424297536 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.958782s. I0416 20:50:53.000133 138701271205440 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:50:53.000254 138701271205440 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:50:53.000344 138701271205440 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:50:53.000391 138701271205440 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 20:50:53.000532 138808848891008 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0416 20:50:53.000675 138808848891008 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.147867 seconds. If this number is high, consider checkpointing less frequently. I0416 20:50:53.001465 138808848891008 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0416 20:50:53.001768 138808848891008 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5. I0416 20:50:53.174251 138808848891008 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:50:53.174409 138701271205440 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5 I0416 20:50:53.174497 138808848891008 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 20:50:53.852257 138701260719680 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5/items I0416 20:51:04.271622 138808848891008 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.098854s I0416 20:51:04.278202 138808848891008 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.102 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 20:51:04.278276 138808848891008 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.197690s (batch_requests_ready=0.084670s, total_serialization_initiated=11.106569s, others=0.006450s) I0416 20:51:04.278362 138808848891008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.198322s (all_items=0.000023s, per_item={'items': '0.00002265'}, temp_paths=11.198299) I0416 20:51:04.279760 138701518669376 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:51:04.279876 138808848891008 async_checkpointer.py:561] Finished blocking save. Time taken: 11.278363s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5. I0416 20:51:04.280111 138808848891008 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:51:04.280335 138701497697856 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:51:04.280445 138808848891008 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_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': 1776372620.8527799, 'wait_for_prev_duration_secs': 32.14786720275879, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776372653.0014918, 'checkpointer_blocking_duration_secs': 11.278503894805908, 'get_old_steps_start_time': 1776372664.280018, 'get_old_steps_duration_secs': 4.935264587402344e-05, 'checkpoint_manager_blocking_start_time': 1776372620.8527117, 'checkpoint_manager_blocking_duration_secs': 43.42770600318909} I0416 20:51:04.280575 138808848891008 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0416 20:51:04.281192 138808848891008 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.958, Tokens/s/device: 17177.029, total_weights: 16384, loss: 8.323 I0416 20:51:04.291829 138808848891008 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.181, Tokens/s/device: 17361.376, total_weights: 16384, loss: 7.360 I0416 20:51:04.405517 138808848891008 metric_logger.py:185] completed step: 6, seconds: 43.667, TFLOP/s/device: 0.311, Tokens/s/device: 46.900, total_weights: 16384, loss: 6.449 I0416 20:51:04.523987 138808848891008 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1606.233, Tokens/s/device: 242108.996, total_weights: 16384, loss: 5.610 I0416 20:51:04.643022 138808848891008 metric_logger.py:185] completed step: 8, seconds: 0.115, TFLOP/s/device: 117.682, Tokens/s/device: 17738.359, total_weights: 16384, loss: 4.852 I0416 20:51:04.761994 138808848891008 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 20:51:04.763351 138808848891008 checkpointing.py:798] Waited 0.001382589340209961 seconds for step 9 to finish before starting checkpointing. I0416 20:51:04.764212 138808848891008 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:51:05.072258 138701529155136 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5/items/array_metadatas/process_0 I0416 20:51:43.094421 138701292176960 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.815580s (commit=38.359103s, array_metadata_write=0.456476s) I0416 20:51:43.095889 138701518669376 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 252.659 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0416 20:51:43.096021 138701518669376 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.816096s. I0416 20:51:43.539626 138701518669376 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 20:51:43.942427 138701518669376 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:51:43.943117 138701518669376 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.562979s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5/items I0416 20:51:43.943860 138701518669376 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5/items I0416 20:51:44.203588 138701518669376 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5 I0416 20:51:44.913635 138701518669376 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5`. I0416 20:51:44.914421 138701518669376 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 51.912915s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/5 I0416 20:51:44.914508 138701518669376 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.634584s. I0416 20:51:44.914796 138701497697856 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:51:44.914929 138701497697856 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:51:44.915005 138701497697856 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:51:44.915068 138701497697856 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0416 20:51:44.915277 138808848891008 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0416 20:51:44.915421 138808848891008 checkpoint_manager.py:1441] Waiting for previous save to complete took 40.151214 seconds. If this number is high, consider checkpointing less frequently. I0416 20:51:44.915680 138808848891008 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 20:51:44.916012 138808848891008 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9. I0416 20:51:45.086387 138701497697856 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9 I0416 20:51:45.108690 138808848891008 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:51:45.108801 138808848891008 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 20:51:45.859087 138701508183616 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9/items I0416 20:51:51.278304 138808848891008 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.170768s I0416 20:51:51.285619 138808848891008 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.964 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 20:51:51.285682 138808848891008 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.282986s (batch_requests_ready=0.096660s, total_serialization_initiated=6.179140s, others=0.007186s) I0416 20:51:51.285767 138808848891008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.283603s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=6.283587) I0416 20:51:51.287255 138701497697856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:51:51.287344 138808848891008 async_checkpointer.py:561] Finished blocking save. Time taken: 6.371624s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9. I0416 20:51:51.287523 138808848891008 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:51:51.287753 138701518669376 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:51:51.287886 138808848891008 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_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': 1776372664.7641816, 'wait_for_prev_duration_secs': 40.15121388435364, 'time_between_consecutive_saves_sec': 11.763775825500488, 'checkpointer_blocking_start_time': 1776372704.915703, 'checkpointer_blocking_duration_secs': 6.3717451095581055, 'get_old_steps_start_time': 1776372711.287461, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776372664.764147, 'checkpoint_manager_blocking_duration_secs': 46.52371335029602} I0416 20:51:51.288033 138808848891008 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 20:51:51.288083 138808848891008 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:51:52.037435 138701529155136 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9/items/array_metadatas/process_0 I0416 20:52:36.187149 138701424297536 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 44.900623s (commit=44.465961s, array_metadata_write=0.434662s) I0416 20:52:36.188451 138701497697856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 246.881 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 51 seconds) (per-host) I0416 20:52:36.188509 138701497697856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.901122s. I0416 20:52:36.592361 138701497697856 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 20:52:36.987345 138701497697856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:52:36.988095 138701497697856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.534090s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9/items I0416 20:52:36.988827 138701497697856 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9/items I0416 20:52:37.233963 138701497697856 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9 I0416 20:52:37.908442 138701497697856 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9`. I0416 20:52:37.909214 138701497697856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.993499s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/checkpoints/9 I0416 20:52:37.909283 138701497697856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 46.621897s. I0416 20:52:37.909467 138701518669376 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:52:37.909584 138701518669376 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:52:37.909641 138701518669376 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:52:37.909681 138701518669376 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 20:52:37.909980 138808848891008 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0416 20:52:37.911472 138808848891008 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:52:37.912804 138808848891008 metric_logger.py:185] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.322, Tokens/s/device: 17382.595, total_weights: 16384, loss: 4.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops