2026-04-16 19:23:12.031892: 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 19:23:12.147788 131331667360896 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 19:24:09.322266 131331667360896 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 19:24:09.322378 131331667360896 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 19:24:09.322416 131331667360896 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 19:24:09.322442 131331667360896 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 19:24:09.322465 131331667360896 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 19:24:09.322547 131331667360896 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 19:24:09.415760 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:09.819593 131331667360896 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:24:09.819761 131331667360896 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:24:09.819970 131331667360896 checkpointing.py:688] Setting up checkpoint logger... I0416 19:24:09.820008 131331667360896 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 19:24:09.820040 131331667360896 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 19:24:09.820698 131331667360896 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 0x7771727f8e90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 19:24:12.221755 131331667360896 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 19:24:12.221919 131331667360896 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 0x776af45943e0>}, handler_registry=None I0416 19:24:12.222170 131331667360896 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x776af45943e0>` 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 19:24:12.222212 131331667360896 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x776af3d3a4b0>` 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 19:24:12.222243 131331667360896 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 0x776af45943e0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x776af45943e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x776af3d3a4b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x776af3d3a4b0>}). I0416 19:24:12.222538 131331667360896 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 19:24:12.222593 131331667360896 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x776af3d31ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 19:24:12.356485 131331667360896 checkpoint_manager.py:1788] Found 0 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 19:24:12.356703 131331667360896 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 0x776af3d39550> I0416 19:24:12.356780 131331667360896 checkpointing.py:302] Checkpoint manager created! W0416 19:24:12.372902 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:12.557636 131331667360896 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 19:24:12.557820 131331667360896 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:24:12.647080 131331667360896 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:24:12.647158 131331667360896 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:24:12.661342 131331667360896 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:24:12.661389 131331667360896 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:24:12.682470 131331667360896 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 19:24:12.682526 131331667360896 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:24:12.696631 131331667360896 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 19:24:12.696686 131331667360896 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:24:12.710865 131331667360896 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 19:24:12.710913 131331667360896 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:24:12.732484 131331667360896 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 19:24:12.732547 131331667360896 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:24:12.750465 131331667360896 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 19:24:12.750515 131331667360896 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 19:24:12.938224 131331667360896 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 19:24:12.938319 131331667360896 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 19:24:12.988701 4033708 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 19:24:13.365292 131331667360896 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 19:24:13.365396 131331667360896 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 19:24:13.365436 131331667360896 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 19:24:13.365480 131331667360896 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 19:24:13.365519 131331667360896 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 19:24:13.365546 131331667360896 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 19:24:13.365585 131331667360896 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 19:24:13.365631 131331667360896 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 19:24:13.365658 131331667360896 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 19:24:13.365684 131331667360896 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 19:24:13.365718 131331667360896 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 19:24:13.365751 131331667360896 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 19:24:13.785005 131331667360896 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:24:13.785097 131331667360896 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 19:24:13.798829 131331667360896 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:24:13.798878 131331667360896 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 19:24:14.425924 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:14.550300 131331667360896 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 19:24:14.550754 131331667360896 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 19:24:14.550898 131331667360896 metric_logger.py:289] number parameters: 1.104 billion W0416 19:24:15.421432 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:15.520009 131331667360896 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 19:24:15.637547 131331667360896 checkpointing.py:798] Waited 0.11753511428833008 seconds for step 0 to finish before starting checkpointing. I0416 19:24:15.638016 131331667360896 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:24:15.638199 131331667360896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 19:24:15.638580 131331667360896 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/0. I0416 19:24:15.718388 131331667360896 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 19:24:15.810826 131331667360896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:24:15.810971 131224102241856 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/0 I0416 19:24:15.811366 131331667360896 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 19:24:16.023468 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.035587 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.046233 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.061978 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.074028 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.080237 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.086649 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.090937 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:16.449518 131224091756096 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/0/items I0416 19:24:16.564761 131331667360896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.754630s I0416 19:24:16.565204 131331667360896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 14.595 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 845 milliseconds) (per-host) I0416 19:24:16.565265 131331667360896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.845677s (batch_requests_ready=0.086822s, total_serialization_initiated=0.758564s, others=0.000292s) I0416 19:24:16.565336 131331667360896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.846503s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.846483) I0416 19:24:16.566216 131224039327296 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:24:16.566337 131331667360896 async_checkpointer.py:561] Finished blocking save. Time taken: 0.928092s. 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/0. I0416 19:24:16.566580 131331667360896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:24:16.566775 131331667360896 standard_logger.py:34] {'step': 0, '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': 1776367455.6380038, 'wait_for_prev_duration_secs': 7.128715515136719e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367455.638222, 'checkpointer_blocking_duration_secs': 0.9282567501068115, 'get_old_steps_start_time': 1776367456.5665057, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776367455.6379142, 'checkpoint_manager_blocking_duration_secs': 0.9288322925567627} I0416 19:24:16.566859 131331667360896 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 19:24:16.566906 131331667360896 max_utils.py:750] Memstats: After params initialized: I0416 19:24:16.567001 131331667360896 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 19:24:16.567028 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 19:24:16.567068 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 19:24:16.567087 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 19:24:16.567104 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 19:24:16.567122 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 19:24:16.567139 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 19:24:16.567155 131331667360896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 19:24:16.573082 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.578372 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:24:16.583059 4033708 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:24:16.607607 131224060298816 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367456361303506, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776367456361303506, "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/0/_CHECKPOINT_METADATA I0416 19:24:16.607830 131224112727616 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:24:16.771159 131331667360896 metric_logger.py:185] completed step: 0, seconds: 0.969, TFLOP/s/device: 14.027, Tokens/s/device: 2114.320, total_weights: 16384, loss: 10.874 I0416 19:24:16.772033 131331667360896 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_16_zarr3_true/tensorboard/' I0416 19:24:16.887475 131331667360896 metric_logger.py:185] completed step: 1, seconds: 1.250, TFLOP/s/device: 10.874, Tokens/s/device: 1639.052, total_weights: 16384, loss: 10.874 I0416 19:24:16.905487 4035888 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 19:24:16.905551 4035888 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 19:24:17.006543 131331667360896 metric_logger.py:185] completed step: 2, seconds: 0.005, TFLOP/s/device: 2549.658, Tokens/s/device: 384312.254, total_weights: 16384, loss: 10.276 I0416 19:24:17.125109 131331667360896 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.953, Tokens/s/device: 17477.684, total_weights: 16384, loss: 9.282 I0416 19:24:17.129554 131331667360896 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0416 19:24:17.361871 131331667360896 checkpointing.py:798] Waited 0.23228669166564941 seconds for step 5 to finish before starting checkpointing. I0416 19:24:17.362248 131331667360896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:24:17.649871 131224102241856 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/0/items/array_metadatas/process_0 I0416 19:24:48.271307 131224049813056 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.363075s (commit=30.916665s, array_metadata_write=0.446410s) I0416 19:24:48.272368 131224039327296 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 388.194 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 32 seconds) (per-host) I0416 19:24:48.272476 131224039327296 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.706084s. I0416 19:24:48.517724 131224039327296 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367456361303506, 'commit_timestamp_nsecs': None, '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/0/_CHECKPOINT_METADATA I0416 19:24:48.711432 131224039327296 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 19:24:48.898842 131224060298816 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367456361303506, '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/0/_CHECKPOINT_METADATA I0416 19:24:49.129952 131224039327296 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:24:49.130140 131224039327296 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570659s. 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/0/items I0416 19:24:49.130748 131224039327296 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/0/items I0416 19:24:49.368764 131224039327296 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/0 I0416 19:24:50.077160 131224039327296 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/0`. I0416 19:24:50.077797 131224039327296 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 34.439563s. 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/0 I0416 19:24:50.077872 131224039327296 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.511481s. I0416 19:24:50.077987 131224112727616 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:24:50.078039 131224112727616 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:24:50.078110 131224112727616 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:24:50.078151 131224112727616 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 19:24:50.078248 131331667360896 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 19:24:50.078312 131331667360896 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.716069 seconds. If this number is high, consider checkpointing less frequently. I0416 19:24:50.079037 131331667360896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0416 19:24:50.079282 131331667360896 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/5. I0416 19:24:50.242924 131224112727616 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/5 I0416 19:24:50.250712 131331667360896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:24:50.250803 131331667360896 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 19:24:50.933454 131223968024128 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/5/items I0416 19:25:01.538216 131331667360896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.287886s I0416 19:25:01.542218 131331667360896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.084 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 19:25:01.542300 131331667360896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.382044s (batch_requests_ready=0.086416s, total_serialization_initiated=11.291724s, others=0.003905s) I0416 19:25:01.542381 131331667360896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.382938s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=11.382920) I0416 19:25:01.543829 131224049813056 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:25:01.543909 131331667360896 async_checkpointer.py:561] Finished blocking save. Time taken: 11.464816s. 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/5. I0416 19:25:01.544120 131331667360896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:25:01.544322 131224039327296 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:25:01.544431 131331667360896 standard_logger.py:34] {'step': 5, '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': 1776367457.3622253, 'wait_for_prev_duration_secs': 32.716068983078, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367490.0790741, 'checkpointer_blocking_duration_secs': 11.464946031570435, 'get_old_steps_start_time': 1776367501.5440404, 'get_old_steps_duration_secs': 3.910064697265625e-05, 'checkpoint_manager_blocking_start_time': 1776367457.362167, 'checkpoint_manager_blocking_duration_secs': 44.182239055633545} I0416 19:25:01.544588 131331667360896 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0416 19:25:01.545063 131331667360896 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.160, Tokens/s/device: 17207.481, total_weights: 16384, loss: 8.327 I0416 19:25:01.552532 131331667360896 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 114.758, Tokens/s/device: 17297.589, total_weights: 16384, loss: 7.367 I0416 19:25:01.668610 131331667360896 metric_logger.py:185] completed step: 6, seconds: 44.421, TFLOP/s/device: 0.306, Tokens/s/device: 46.104, total_weights: 16384, loss: 6.459 I0416 19:25:01.787013 131331667360896 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2404.802, Tokens/s/device: 362477.876, total_weights: 16384, loss: 5.622 I0416 19:25:01.905447 131331667360896 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.216, Tokens/s/device: 17517.299, total_weights: 16384, loss: 4.864 I0416 19:25:02.023405 131331667360896 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 19:25:02.024182 131331667360896 checkpointing.py:798] Waited 0.0008034706115722656 seconds for step 9 to finish before starting checkpointing. I0416 19:25:02.024446 131331667360896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:25:02.312520 131224102241856 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/5/items/array_metadatas/process_0 I0416 19:25:40.340651 131223968024128 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.797606s (commit=38.232588s, array_metadata_write=0.565017s) I0416 19:25:40.341804 131224049813056 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 251.821 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0416 19:25:40.341860 131224049813056 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.797908s. I0416 19:25:40.760882 131224049813056 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 19:25:41.177950 131224049813056 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:25:41.178630 131224049813056 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.555868s. 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/5/items I0416 19:25:41.179286 131224049813056 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/5/items I0416 19:25:41.410221 131224049813056 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/5 I0416 19:25:42.081517 131224049813056 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/5`. I0416 19:25:42.082189 131224049813056 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.003103s. 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/5 I0416 19:25:42.082262 131224049813056 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.538310s. I0416 19:25:42.082489 131224039327296 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:25:42.082596 131224039327296 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:25:42.082666 131224039327296 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:25:42.082707 131224039327296 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0416 19:25:42.082888 131331667360896 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 19:25:42.083010 131331667360896 checkpoint_manager.py:1441] Waiting for previous save to complete took 40.058567 seconds. If this number is high, consider checkpointing less frequently. I0416 19:25:42.083267 131331667360896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 19:25:42.083558 131331667360896 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/9. I0416 19:25:42.255129 131224039327296 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/9 I0416 19:25:42.260612 131331667360896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:25:42.260703 131331667360896 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 19:25:42.949704 131223999481408 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/9/items I0416 19:25:53.519470 131331667360896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.259249s I0416 19:25:53.523609 131331667360896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.087 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 19:25:53.523664 131331667360896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.351509s (batch_requests_ready=0.084429s, total_serialization_initiated=11.263050s, others=0.004030s) I0416 19:25:53.523739 131331667360896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.352019s (all_items=0.000014s, per_item={'items': '0.00001359'}, temp_paths=11.352005) I0416 19:25:53.525208 131224102241856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:25:53.525291 131331667360896 async_checkpointer.py:561] Finished blocking save. Time taken: 11.441979s. 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/9. I0416 19:25:53.525488 131331667360896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:25:53.525737 131223978509888 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:25:53.525854 131331667360896 standard_logger.py:34] {'step': 9, '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': 1776367502.0244224, 'wait_for_prev_duration_secs': 40.05856680870056, 'time_between_consecutive_saves_sec': 11.946256875991821, 'checkpointer_blocking_start_time': 1776367542.0832922, 'checkpointer_blocking_duration_secs': 11.44210147857666, 'get_old_steps_start_time': 1776367553.5254147, 'get_old_steps_duration_secs': 3.5762786865234375e-05, 'checkpoint_manager_blocking_start_time': 1776367502.0243914, 'checkpoint_manager_blocking_duration_secs': 51.50143647193909} I0416 19:25:53.525977 131331667360896 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 19:25:53.526008 131331667360896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:25:54.240067 131224112727616 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/9/items/array_metadatas/process_0 I0416 19:26:36.922235 131224039327296 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 43.398021s (commit=42.954273s, array_metadata_write=0.443749s) I0416 19:26:36.923228 131224102241856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 230.804 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 54 seconds) (per-host) I0416 19:26:36.923284 131224102241856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.397947s. I0416 19:26:37.349475 131224102241856 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 19:26:37.767303 131224102241856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 19:26:37.767469 131224102241856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557851s. 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/9/items I0416 19:26:37.768091 131224102241856 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/9/items I0416 19:26:38.009639 131224102241856 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/9 I0416 19:26:38.680208 131224102241856 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/9`. I0416 19:26:38.680828 131224102241856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 56.597523s. 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/9 I0416 19:26:38.680894 131224102241856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 45.155561s. I0416 19:26:38.681062 131223978509888 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:26:38.681166 131223978509888 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:26:38.681229 131223978509888 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:26:38.681277 131223978509888 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 19:26:38.681414 131331667360896 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 19:26:38.681550 131331667360896 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:26:38.682230 131331667360896 metric_logger.py:185] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.791, Tokens/s/device: 17302.558, total_weights: 16384, loss: 4.189 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops