2026-04-16 19:18:11.843625: 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 19:18:11.959211 136761269365888 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 19:19:10.775261 136761269365888 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 19:19:10.775373 136761269365888 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 19:19:10.775409 136761269365888 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 19:19:10.775434 136761269365888 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 19:19:10.775455 136761269365888 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 19:19:10.775531 136761269365888 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 19:19:10.867359 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:11.263134 136761269365888 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:19:11.263301 136761269365888 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 19:19:11.263461 136761269365888 checkpointing.py:688] Setting up checkpoint logger... I0416 19:19:11.263499 136761269365888 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0416 19:19:11.263532 136761269365888 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 19:19:11.264010 136761269365888 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7c619fff8dd0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 19:19:13.666518 136761269365888 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 19:19:13.666705 136761269365888 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 0x7c5b24f73380>}, handler_registry=None I0416 19:19:13.666947 136761269365888 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c5b24f73380>` 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 19:19:13.666987 136761269365888 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5b23d39a30>` 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 19:19:13.667017 136761269365888 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 0x7c5b24f73380>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c5b24f73380>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5b23d39a30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5b23d39a30>}). I0416 19:19:13.667579 136761269365888 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 19:19:13.667636 136761269365888 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7c5b23d31ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 19:19:13.792761 136761269365888 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_15_ocdbt_false/checkpoints I0416 19:19:13.792977 136761269365888 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_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c5b23d399a0> I0416 19:19:13.793074 136761269365888 checkpointing.py:302] Checkpoint manager created! W0416 19:19:13.809512 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:13.996420 136761269365888 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 19:19:13.996513 136761269365888 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:19:14.093512 136761269365888 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:19:14.093676 136761269365888 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:19:14.107931 136761269365888 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 19:19:14.107980 136761269365888 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 19:19:14.130070 136761269365888 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 19:19:14.130128 136761269365888 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:19:14.144249 136761269365888 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 19:19:14.144296 136761269365888 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:19:14.158389 136761269365888 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 19:19:14.158435 136761269365888 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:19:14.180723 136761269365888 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 19:19:14.180780 136761269365888 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 19:19:14.198628 136761269365888 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 19:19:14.198678 136761269365888 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 19:19:14.407627 136761269365888 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 19:19:14.407744 136761269365888 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 19:19:14.458669 4028637 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 19:19:14.838422 136761269365888 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 19:19:14.838530 136761269365888 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 19:19:14.838570 136761269365888 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 19:19:14.838620 136761269365888 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 19:19:14.838659 136761269365888 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 19:19:14.838687 136761269365888 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 19:19:14.838726 136761269365888 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 19:19:14.838764 136761269365888 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 19:19:14.838791 136761269365888 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 19:19:14.838815 136761269365888 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 19:19:14.838849 136761269365888 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 19:19:14.838881 136761269365888 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 19:19:15.261243 136761269365888 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:19:15.261323 136761269365888 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 19:19:15.274991 136761269365888 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 19:19:15.275041 136761269365888 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 19:19:15.922727 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:16.049341 136761269365888 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 19:19:16.049817 136761269365888 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 19:19:16.049958 136761269365888 metric_logger.py:289] number parameters: 1.104 billion W0416 19:19:16.919412 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:17.018648 136761269365888 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 19:19:17.135768 136761269365888 checkpointing.py:798] Waited 0.11710739135742188 seconds for step 0 to finish before starting checkpointing. I0416 19:19:17.136214 136761269365888 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:19:17.136362 136761269365888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 19:19:17.136711 136761269365888 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_15_ocdbt_false/checkpoints/0. I0416 19:19:17.225180 136761269365888 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 19:19:17.312177 136761269365888 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:19:17.312266 136761269365888 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 19:19:17.321922 136653714753088 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_15_ocdbt_false/checkpoints/0 W0416 19:19:17.525865 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.536231 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.549245 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.568744 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.579771 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.586260 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.590631 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.594555 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:17.783140 136761269365888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.471366s I0416 19:19:17.783690 136761269365888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 22.150 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 557 milliseconds) (per-host) I0416 19:19:17.783750 136761269365888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.557253s (batch_requests_ready=0.081643s, total_serialization_initiated=0.475152s, others=0.000458s) I0416 19:19:17.783819 136761269365888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.558189s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.558169) I0416 19:19:17.784657 136653651838528 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:19:17.784753 136761269365888 async_checkpointer.py:561] Finished blocking save. Time taken: 0.648354s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/0. I0416 19:19:17.784949 136761269365888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:19:17.785125 136653725238848 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:19:17.785230 136761269365888 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367157.1362023, 'wait_for_prev_duration_secs': 4.2438507080078125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367157.1363816, 'checkpointer_blocking_duration_secs': 0.6484935283660889, 'get_old_steps_start_time': 1776367157.7848911, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776367157.136121, 'checkpoint_manager_blocking_duration_secs': 0.6490797996520996} I0416 19:19:17.785351 136761269365888 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 19:19:17.785397 136761269365888 max_utils.py:750] Memstats: After params initialized: I0416 19:19:17.785485 136761269365888 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 19:19:17.785519 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 19:19:17.785546 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 19:19:17.785571 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 19:19:17.785589 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 19:19:17.785605 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 19:19:17.785620 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 19:19:17.785635 136761269365888 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 19:19:17.790989 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.796198 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 19:19:17.800765 4028637 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 19:19:17.977503 136653704267328 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_15_ocdbt_false/checkpoints/0/items I0416 19:19:17.982924 136761269365888 metric_logger.py:185] completed step: 0, seconds: 0.968, TFLOP/s/device: 14.033, Tokens/s/device: 2115.211, total_weights: 16384, loss: 10.874 I0416 19:19:17.983705 136761269365888 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_15_ocdbt_false/tensorboard/' I0416 19:19:18.099233 136761269365888 metric_logger.py:185] completed step: 1, seconds: 0.963, TFLOP/s/device: 14.114, Tokens/s/device: 2127.350, total_weights: 16384, loss: 10.874 I0416 19:19:18.099449 136653632964160 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367157872066164, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776367157872066164, "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_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0416 19:19:18.218208 136761269365888 metric_logger.py:185] completed step: 2, seconds: 0.005, TFLOP/s/device: 2625.025, Tokens/s/device: 395672.334, total_weights: 16384, loss: 10.276 I0416 19:19:18.336724 136761269365888 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.648, Tokens/s/device: 17431.716, total_weights: 16384, loss: 9.282 I0416 19:19:18.341311 136761269365888 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0416 19:19:18.438628 4030802 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 19:19:18.438680 4030802 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 19:19:18.573871 136761269365888 checkpointing.py:798] Waited 0.23253321647644043 seconds for step 5 to finish before starting checkpointing. I0416 19:19:18.574261 136761269365888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:19:19.167287 136653672810048 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_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_0 I0416 19:19:24.679799 136653662324288 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.239508s (commit=5.816571s, array_metadata_write=0.422937s) I0416 19:19:24.680819 136653651838528 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1.656 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0416 19:19:24.680926 136653651838528 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.896122s. I0416 19:19:24.908200 136653651838528 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367157872066164, '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_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA W0416 19:19:24.993810 136653651838528 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/0/items, I0416 19:19:25.071178 136653651838528 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 19:19:25.071280 136653651838528 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.130312s. 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_15_ocdbt_false/checkpoints/0/items I0416 19:19:25.071914 136653651838528 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/0/items I0416 19:19:25.279468 136653632964160 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776367157872066164, '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_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0416 19:19:25.297733 136653651838528 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/0 I0416 19:19:25.953879 136653651838528 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_15_ocdbt_false/checkpoints/0`. I0416 19:19:25.954506 136653651838528 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 8.818113s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/0 I0416 19:19:25.954580 136653651838528 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 8.169778s. I0416 19:19:25.954761 136653725238848 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:19:25.954870 136653725238848 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:19:25.954937 136653725238848 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:19:25.954983 136653725238848 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 19:19:25.955146 136761269365888 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 19:19:25.955263 136761269365888 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.381006 seconds. If this number is high, consider checkpointing less frequently. I0416 19:19:25.955982 136761269365888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0416 19:19:25.956282 136761269365888 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_15_ocdbt_false/checkpoints/5. I0416 19:19:26.129205 136761269365888 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:19:26.129346 136653725238848 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_15_ocdbt_false/checkpoints/5 I0416 19:19:26.129424 136761269365888 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 19:19:26.798878 136653662324288 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_15_ocdbt_false/checkpoints/5/items I0416 19:19:36.747423 136761269365888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.618934s I0416 19:19:36.751663 136761269365888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.153 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0416 19:19:36.751722 136761269365888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.705801s (batch_requests_ready=0.079830s, total_serialization_initiated=10.621839s, others=0.004133s) I0416 19:19:36.751805 136761269365888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.706325s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=10.706308) I0416 19:19:36.753138 136653580535360 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:19:36.753229 136761269365888 async_checkpointer.py:561] Finished blocking save. Time taken: 10.797195s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/5. I0416 19:19:36.753477 136761269365888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:19:36.753637 136653651838528 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:19:36.753717 136761269365888 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367158.5742393, 'wait_for_prev_duration_secs': 7.381006479263306, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776367165.9560049, 'checkpointer_blocking_duration_secs': 10.797364234924316, 'get_old_steps_start_time': 1776367176.7533956, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776367158.5741765, 'checkpoint_manager_blocking_duration_secs': 18.179513216018677} I0416 19:19:36.753843 136761269365888 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0416 19:19:36.754364 136761269365888 metric_logger.py:185] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.567, Tokens/s/device: 17268.711, total_weights: 16384, loss: 8.327 I0416 19:19:36.761703 136761269365888 metric_logger.py:185] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.591, Tokens/s/device: 17272.352, total_weights: 16384, loss: 7.367 I0416 19:19:36.877520 136761269365888 metric_logger.py:185] completed step: 6, seconds: 18.419, TFLOP/s/device: 0.738, Tokens/s/device: 111.192, total_weights: 16384, loss: 6.459 I0416 19:19:36.999218 136761269365888 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 2503.156, Tokens/s/device: 377302.874, total_weights: 16384, loss: 5.622 I0416 19:19:37.116015 136761269365888 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.091, Tokens/s/device: 17498.441, total_weights: 16384, loss: 4.864 I0416 19:19:37.235839 136761269365888 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 19:19:37.237172 136761269365888 checkpointing.py:798] Waited 0.001359701156616211 seconds for step 9 to finish before starting checkpointing. I0416 19:19:37.237510 136761269365888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:19:37.571326 136653672810048 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_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_0 I0416 19:19:43.432349 136653662324288 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.679976s (commit=6.181932s, array_metadata_write=0.498044s) I0416 19:19:43.433349 136653580535360 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 726.779 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 17 seconds) (per-host) I0416 19:19:43.433405 136653580535360 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.680119s. W0416 19:19:43.732118 136653580535360 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/5/items, I0416 19:19:43.832336 136653580535360 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 19:19:43.832446 136653580535360 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.152316s. 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_15_ocdbt_false/checkpoints/5/items I0416 19:19:43.833129 136653580535360 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/5/items I0416 19:19:44.073973 136653580535360 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/5 I0416 19:19:44.777221 136653580535360 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_15_ocdbt_false/checkpoints/5`. I0416 19:19:44.777874 136653580535360 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 18.821857s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/5 I0416 19:19:44.777954 136653580535360 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 8.024668s. I0416 19:19:44.778136 136653651838528 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:19:44.778236 136653651838528 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:19:44.778304 136653651838528 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:19:44.778348 136653651838528 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0416 19:19:44.778496 136761269365888 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 19:19:44.778615 136761269365888 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.541117 seconds. If this number is high, consider checkpointing less frequently. I0416 19:19:44.778837 136761269365888 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 19:19:44.779136 136761269365888 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_15_ocdbt_false/checkpoints/9. I0416 19:19:44.926456 136653580535360 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_15_ocdbt_false/checkpoints/9 I0416 19:19:44.965697 136761269365888 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 19:19:44.965793 136761269365888 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 19:19:45.599405 136653662324288 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_15_ocdbt_false/checkpoints/9/items I0416 19:19:54.665835 136761269365888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.700490s I0416 19:19:54.669845 136761269365888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.258 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0416 19:19:54.669906 136761269365888 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.808897s (batch_requests_ready=0.100741s, total_serialization_initiated=9.704247s, others=0.003908s) I0416 19:19:54.669981 136761269365888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.809423s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=9.809409) I0416 19:19:54.671279 136653591021120 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 19:19:54.671487 136761269365888 async_checkpointer.py:561] Finished blocking save. Time taken: 9.892610s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9. I0416 19:19:54.671639 136761269365888 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 19:19:54.671839 136651804247616 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 19:19:54.671940 136761269365888 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776367177.2374828, 'wait_for_prev_duration_secs': 7.541116714477539, 'time_between_consecutive_saves_sec': 11.282484292984009, 'checkpointer_blocking_start_time': 1776367184.77886, 'checkpointer_blocking_duration_secs': 9.892704010009766, 'get_old_steps_start_time': 1776367194.671579, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776367177.2374444, 'checkpoint_manager_blocking_duration_secs': 17.43447184562683} I0416 19:19:54.672083 136761269365888 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 19:19:54.672116 136761269365888 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 19:19:55.391476 136653672810048 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_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_0 I0416 19:20:01.520645 136653662324288 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.850077s (commit=6.428472s, array_metadata_write=0.421604s) I0416 19:20:01.521633 136653591021120 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 758.485 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 16 seconds) (per-host) I0416 19:20:01.521740 136653591021120 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.850288s. W0416 19:20:01.830516 136653591021120 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9/items, I0416 19:20:01.921359 136653591021120 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 19:20:01.921476 136653591021120 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.151053s. 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_15_ocdbt_false/checkpoints/9/items I0416 19:20:01.922117 136653591021120 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9/items I0416 19:20:02.153460 136653591021120 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9 I0416 19:20:02.845166 136653591021120 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_15_ocdbt_false/checkpoints/9`. I0416 19:20:02.845806 136653591021120 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 18.066935s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_15_ocdbt_false/checkpoints/9 I0416 19:20:02.845878 136653591021120 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 8.174427s. I0416 19:20:02.846078 136651804247616 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 19:20:02.846187 136651804247616 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 19:20:02.846234 136651804247616 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 19:20:02.846270 136651804247616 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 19:20:02.846435 136761269365888 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 19:20:02.846569 136761269365888 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 19:20:02.847248 136761269365888 metric_logger.py:185] completed step: 9, seconds: 0.124, TFLOP/s/device: 109.431, Tokens/s/device: 16494.580, 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