2026-04-23 04:04:41.692407: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776917081.706250 2408209 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776917081.710209 2408209 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776917081.722688 2408209 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776917081.722704 2408209 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776917081.722706 2408209 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776917081.722707 2408209 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-23 04:04:45.375158: 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) I0423 04:04:45.512403 123973702715200 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. ~/maxtext/maxtext_venv/lib/python3.12/site-packages/jax/_src/xla_bridge.py:204: 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( I0423 04:05:47.596722 123973702715200 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 04:05:47.596961 123973702715200 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 04:05:47.597016 123973702715200 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0423 04:05:47.597057 123973702715200 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 04:05:47.597086 123973702715200 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 04:05:47.597242 123973702715200 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 04:05:48.260026 123973702715200 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 04:05:48.260295 123973702715200 checkpointing.py:677] Setting up checkpoint logger... I0423 04:05:48.260341 123973702715200 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0423 04:05:48.260377 123973702715200 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 04:05:48.260798 123973702715200 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x70c096f9f110>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 04:05:50.660350 123973702715200 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 04:05:50.660707 123973702715200 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 0x70bf3af87bf0>}, handler_registry=None I0423 04:05:50.660923 123973702715200 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70bf3af87bf0>` 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`. I0423 04:05:50.660965 123973702715200 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70bebaf1ec60>` 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`. I0423 04:05:50.660993 123973702715200 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 0x70bf3af87bf0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70bf3af87bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70bebaf1ec60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70bebaf1ec60>}). I0423 04:05:50.661297 123973702715200 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 04:05:50.661348 123973702715200 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x70bebaf50540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 04:05:50.781022 123973702715200 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints I0423 04:05:50.781262 123973702715200 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_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x70bebaf1e6c0> I0423 04:05:50.781543 123973702715200 checkpointing.py:301] Checkpoint manager created! I0423 04:05:51.187202 123973702715200 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 04:05:51.187292 123973702715200 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 04:05:51.466474 123973702715200 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 04:05:51.466556 123973702715200 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 04:05:51.478748 123973702715200 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 04:05:51.478796 123973702715200 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 04:05:51.497633 123973702715200 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 04:05:51.497689 123973702715200 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 04:05:51.509900 123973702715200 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 04:05:51.509948 123973702715200 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 04:05:51.522089 123973702715200 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 04:05:51.522140 123973702715200 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 04:05:51.541345 123973702715200 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 04:05:51.541402 123973702715200 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 04:05:51.556607 123973702715200 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 04:05:51.556660 123973702715200 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 04:05:51.709118 123973702715200 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 04:05:51.709211 123973702715200 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [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 I0423 04:05:52.346545 123973702715200 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 04:05:52.346666 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 04:05:52.346709 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 04:05:52.346756 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0423 04:05:52.346800 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 04:05:52.346828 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 04:05:52.346868 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 04:05:52.346908 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0423 04:05:52.346935 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0423 04:05:52.346961 123973702715200 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 04:05:52.346994 123973702715200 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 04:05:52.347029 123973702715200 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 04:05:52.750020 123973702715200 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 04:05:52.750114 123973702715200 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 04:05:52.761646 123973702715200 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 04:05:52.761695 123973702715200 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 04:05:53.525667 123973702715200 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. I0423 04:05:53.526286 123973702715200 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 04:05:53.526432 123973702715200 metric_logger.py:301] number parameters: 1.104 billion I0423 04:05:55.985946 123973702715200 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 04:05:56.104112 123973702715200 checkpointing.py:776] Waited 0.11815166473388672 seconds for step 0 to finish before starting checkpointing. I0423 04:05:56.104704 123973702715200 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 04:05:56.104860 123973702715200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 04:05:56.105226 123973702715200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0. I0423 04:05:56.188633 123973702715200 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 04:05:56.273841 123942352193088 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0 I0423 04:05:56.280149 123973702715200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 04:05:56.280238 123973702715200 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 I0423 04:05:57.015289 123942341707328 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/items I0423 04:05:57.187096 123942310250048 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776917156924988442, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776917156924988442, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0423 04:05:57.519255 2410381 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 04:05:57.519309 2410381 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 04:05:58.081015 123973702715200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.801255s I0423 04:05:58.085393 123973702715200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.511 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 04:05:58.085456 123973702715200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.895468s (batch_requests_ready=0.085959s, total_serialization_initiated=1.805234s, others=0.004275s) I0423 04:05:58.085543 123973702715200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.896434s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=1.896415) I0423 04:05:58.086783 123941511235136 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 04:05:58.086899 123973702715200 async_checkpointer.py:561] Finished blocking save. Time taken: 1.981999s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0. I0423 04:05:58.087035 123973702715200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 04:05:58.087235 123942498993728 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 04:05:58.087355 123973702715200 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776917156.104691, 'wait_for_prev_duration_secs': 4.1484832763671875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776917156.104882, 'checkpointer_blocking_duration_secs': 1.9820747375488281, 'get_old_steps_start_time': 1776917158.0869753, 'get_old_steps_duration_secs': 2.3365020751953125e-05, 'checkpoint_manager_blocking_start_time': 1776917156.1046128, 'checkpoint_manager_blocking_duration_secs': 1.9827141761779785} I0423 04:05:58.087464 123973702715200 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 04:05:58.087520 123973702715200 max_utils.py:750] Memstats: After params initialized: I0423 04:05:58.087574 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 04:05:58.087599 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 04:05:58.087619 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 04:05:58.087637 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 04:05:58.087654 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 04:05:58.087671 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 04:05:58.087687 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 04:05:58.087704 123973702715200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 04:05:58.348893 123973702715200 metric_logger.py:196] completed step: 0, seconds: 2.459, TFLOP/s/device: 5.525, Tokens/s/device: 832.855, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 04:05:58.349671 123973702715200 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/tensorboard/' I0423 04:05:58.466099 123973702715200 metric_logger.py:196] completed step: 1, seconds: 2.361, TFLOP/s/device: 5.755, Tokens/s/device: 867.436, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 04:05:58.585324 123973702715200 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2339.785, Tokens/s/device: 352677.803, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0423 04:05:58.704676 123973702715200 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 114.785, Tokens/s/device: 17301.681, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0423 04:05:58.709622 123973702715200 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0423 04:05:58.846155 123942320735808 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0423 04:05:58.942010 123973702715200 checkpointing.py:776] Waited 0.23235821723937988 seconds for step 5 to finish before starting checkpointing. I0423 04:05:58.942909 123973702715200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 04:06:35.811195 123941712561728 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.725118s (commit=37.187736s, array_metadata_write=0.537382s) I0423 04:06:35.812223 123941511235136 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 318.932 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0423 04:06:35.812278 123941511235136 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.725415s. I0423 04:06:36.092863 123941511235136 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776917156924988442, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0423 04:06:36.325217 123941511235136 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 04:06:36.521602 123942310250048 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776917156924988442, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0423 04:06:36.882735 123941511235136 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.744271s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/items I0423 04:06:36.883463 123941511235136 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0/items I0423 04:06:37.125549 123941511235136 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0 I0423 04:06:37.847562 123941511235136 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0`. I0423 04:06:37.848209 123941511235136 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.743315s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/0 I0423 04:06:37.848280 123941511235136 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.761418s. I0423 04:06:37.848506 123942498993728 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 04:06:37.848621 123942498993728 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 04:06:37.848687 123942498993728 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 04:06:37.848732 123942498993728 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 04:06:37.848884 123973702715200 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. W0423 04:06:37.849007 123973702715200 checkpoint_manager.py:1441] Waiting for previous save to complete took 38.906099 seconds. If this number is high, consider checkpointing less frequently. I0423 04:06:37.850296 123973702715200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0423 04:06:37.850586 123973702715200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5. I0423 04:06:38.019813 123941511235136 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5 I0423 04:06:38.028772 123973702715200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 04:06:38.028867 123973702715200 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 I0423 04:06:38.797191 123942320735808 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5/items I0423 04:06:40.206694 123973702715200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.178281s I0423 04:06:40.211410 123973702715200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.421 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 04:06:40.211492 123973702715200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.276527s (batch_requests_ready=0.090088s, total_serialization_initiated=2.181885s, others=0.004554s) I0423 04:06:40.211609 123973702715200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.277087s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=2.277071) I0423 04:06:40.213623 123941712561728 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 04:06:40.213768 123973702715200 async_checkpointer.py:561] Finished blocking save. Time taken: 2.363419s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5. I0423 04:06:40.214075 123973702715200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0423 04:06:40.214282 123942498993728 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 04:06:40.214426 123973702715200 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776917158.9428847, 'wait_for_prev_duration_secs': 38.90609884262085, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776917197.85032, 'checkpointer_blocking_duration_secs': 2.363593339920044, 'get_old_steps_start_time': 1776917200.2139344, 'get_old_steps_duration_secs': 4.9114227294921875e-05, 'checkpoint_manager_blocking_start_time': 1776917158.942813, 'checkpoint_manager_blocking_duration_secs': 41.271578311920166} I0423 04:06:40.214620 123973702715200 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0423 04:06:40.215427 123973702715200 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.953, Tokens/s/device: 17176.309, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0423 04:06:40.228858 123973702715200 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.101, Tokens/s/device: 17198.522, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0423 04:06:40.343099 123973702715200 metric_logger.py:196] completed step: 6, seconds: 41.515, TFLOP/s/device: 0.327, Tokens/s/device: 49.331, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0423 04:06:40.461855 123973702715200 metric_logger.py:196] completed step: 7, seconds: 0.010, TFLOP/s/device: 1379.824, Tokens/s/device: 207982.127, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0423 04:06:40.581295 123973702715200 metric_logger.py:196] completed step: 8, seconds: 0.115, TFLOP/s/device: 118.220, Tokens/s/device: 17819.387, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0423 04:06:40.699465 123973702715200 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 04:06:40.700795 123973702715200 checkpointing.py:776] Waited 0.0014011859893798828 seconds for step 9 to finish before starting checkpointing. I0423 04:06:40.701169 123973702715200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 04:06:40.993860 123942150866496 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0423 04:07:13.682082 123942104729152 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.469656s (commit=33.000751s, array_metadata_write=0.468904s) I0423 04:07:13.683708 123941712561728 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 353.490 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 35 seconds) (per-host) I0423 04:07:13.683882 123941712561728 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.470052s. I0423 04:07:14.155812 123941712561728 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 04:07:14.612370 123941712561728 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.622983s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5/items I0423 04:07:14.613660 123941712561728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5/items I0423 04:07:14.866704 123941712561728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5 I0423 04:07:15.576344 123941712561728 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5`. I0423 04:07:15.577276 123941712561728 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 37.726940s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/5 I0423 04:07:15.577382 123941712561728 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.363556s. I0423 04:07:15.577729 123942498993728 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 04:07:15.577897 123942498993728 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 04:07:15.577968 123942498993728 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0423 04:07:15.578017 123942498993728 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0423 04:07:15.578270 123973702715200 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. W0423 04:07:15.578398 123973702715200 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.877250 seconds. If this number is high, consider checkpointing less frequently. I0423 04:07:15.578722 123973702715200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 04:07:15.579032 123973702715200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9. I0423 04:07:15.740854 123941712561728 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9 I0423 04:07:15.752317 123973702715200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 04:07:15.752409 123973702715200 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 I0423 04:07:16.456644 123942150866496 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9/items I0423 04:07:20.183125 123973702715200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.431156s I0423 04:07:20.187085 123973702715200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.727 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0423 04:07:20.187144 123973702715200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.525865s (batch_requests_ready=0.087355s, total_serialization_initiated=4.434654s, others=0.003856s) I0423 04:07:20.187221 123973702715200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.526373s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=4.526361) I0423 04:07:20.188858 123941712561728 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 04:07:20.188942 123973702715200 async_checkpointer.py:561] Finished blocking save. Time taken: 4.610174s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9. I0423 04:07:20.189127 123973702715200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 04:07:20.190162 123942498993728 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 04:07:20.190277 123973702715200 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776917200.701131, 'wait_for_prev_duration_secs': 34.87725019454956, 'time_between_consecutive_saves_sec': 2.852383613586426, 'checkpointer_blocking_start_time': 1776917235.5787485, 'checkpointer_blocking_duration_secs': 4.610316514968872, 'get_old_steps_start_time': 1776917240.1890783, 'get_old_steps_duration_secs': 1.7404556274414062e-05, 'checkpoint_manager_blocking_start_time': 1776917200.701086, 'checkpoint_manager_blocking_duration_secs': 39.4891676902771} I0423 04:07:20.190399 123973702715200 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 04:07:20.190430 123973702715200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 04:07:20.961951 123942320735808 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0423 04:07:55.026886 123942150866496 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.839176s (commit=34.345823s, array_metadata_write=0.493353s) I0423 04:07:55.027822 123941712561728 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 321.003 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0423 04:07:55.027877 123941712561728 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.838892s. I0423 04:07:55.493533 123941712561728 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 04:07:55.972399 123941712561728 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.642746s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9/items I0423 04:07:55.973108 123941712561728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9/items I0423 04:07:56.225596 123941712561728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9 I0423 04:07:56.920326 123941712561728 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9`. I0423 04:07:56.921014 123941712561728 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.342253s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_16_zarr3_false/checkpoints/9 I0423 04:07:56.921092 123941712561728 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.732107s. I0423 04:07:56.921305 123942498993728 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 04:07:56.921415 123942498993728 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 04:07:56.921459 123942498993728 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 04:07:56.921499 123942498993728 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 04:07:56.921638 123973702715200 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. I0423 04:07:56.921769 123973702715200 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 04:07:56.922638 123973702715200 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.369, Tokens/s/device: 17238.912, total_weights: 16384, loss: 4.189, lm_loss: 4.189, perplexity: 65.979 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops