2026-04-23 03:33:08.283899: 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 00:00:1776915188.297517 2382496 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776915188.301523 2382496 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776915188.314024 2382496 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776915188.314038 2382496 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776915188.314040 2382496 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776915188.314053 2382496 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 03:33:12.190240: 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 03:33:12.326550 140238750848832 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 03:34:17.770795 140238750848832 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 03:34:17.770904 140238750848832 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 03:34:17.770938 140238750848832 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 03:34:17.770964 140238750848832 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 03:34:17.770985 140238750848832 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 03:34:17.771077 140238750848832 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 03:34:18.445961 140238750848832 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 03:34:18.446240 140238750848832 checkpointing.py:677] Setting up checkpoint logger... I0423 03:34:18.446287 140238750848832 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 03:34:18.446325 140238750848832 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 03:34:18.446760 140238750848832 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7f8b975a9be0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 03:34:20.947277 140238750848832 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 03:34:20.947649 140238750848832 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 0x7f8a3c7b54c0>}, handler_registry=None I0423 03:34:20.947874 140238750848832 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f8a3c7b54c0>` 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 03:34:20.947915 140238750848832 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f89bab22d20>` 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 03:34:20.947944 140238750848832 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 0x7f8a3c7b54c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f8a3c7b54c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f89bab22d20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f89bab22d20>}). I0423 03:34:20.948252 140238750848832 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 03:34:20.948309 140238750848832 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7f89bab50540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 03:34:21.079262 140238750848832 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_13_scan_layers_true/checkpoints I0423 03:34:21.079478 140238750848832 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=10), 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_13_scan_layers_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f89baf8c080> I0423 03:34:21.079772 140238750848832 checkpointing.py:301] Checkpoint manager created! I0423 03:34:21.473773 140238750848832 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 03:34:21.473862 140238750848832 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:34:21.763689 140238750848832 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:34:21.763798 140238750848832 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:34:21.776167 140238750848832 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:34:21.776220 140238750848832 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:34:21.795850 140238750848832 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 03:34:21.795903 140238750848832 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:34:21.808475 140238750848832 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 03:34:21.808524 140238750848832 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:34:21.820740 140238750848832 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 03:34:21.820791 140238750848832 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:34:21.840384 140238750848832 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 03:34:21.840441 140238750848832 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:34:21.855890 140238750848832 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 03:34:21.855941 140238750848832 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 03:34:22.007842 140238750848832 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 03:34:22.007949 140238750848832 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 03:34:22.657736 140238750848832 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:34:22.657848 140238750848832 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 03:34:22.657891 140238750848832 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 03:34:22.657936 140238750848832 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 03:34:22.657978 140238750848832 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 03:34:22.658006 140238750848832 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 03:34:22.658062 140238750848832 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 03:34:22.658107 140238750848832 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 03:34:22.658137 140238750848832 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 03:34:22.658164 140238750848832 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 03:34:22.658200 140238750848832 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 03:34:22.658235 140238750848832 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 03:34:23.067626 140238750848832 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:34:23.067712 140238750848832 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:34:23.079715 140238750848832 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:34:23.079766 140238750848832 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:34:23.842597 140238750848832 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 03:34:23.843172 140238750848832 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 03:34:23.843339 140238750848832 metric_logger.py:301] number parameters: 1.104 billion I0423 03:34:26.062421 140238750848832 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 03:34:26.180128 140238750848832 checkpointing.py:776] Waited 0.11769604682922363 seconds for step 0 to finish before starting checkpointing. I0423 03:34:26.180755 140238750848832 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:34:26.180912 140238750848832 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 03:34:26.181302 140238750848832 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_13_scan_layers_true/checkpoints/0. I0423 03:34:26.259037 140238750848832 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 03:34:26.339848 140207393343040 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_13_scan_layers_true/checkpoints/0 I0423 03:34:26.352116 140238750848832 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:34:26.352205 140238750848832 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 03:34:27.088149 140207382857280 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_13_scan_layers_true/checkpoints/0/items I0423 03:34:27.254002 140207347205696 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915267006655757, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776915267006655757, "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_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:34:27.564881 2384581 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 03:34:27.564932 2384581 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 03:34:28.185651 140238750848832 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.833940s I0423 03:34:28.190456 140238750848832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.394 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 03:34:28.190522 140238750848832 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.930129s (batch_requests_ready=0.087456s, total_serialization_initiated=1.837959s, others=0.004714s) I0423 03:34:28.190595 140238750848832 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.931112s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=1.931093) I0423 03:34:28.191863 140207542240832 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:34:28.191950 140238750848832 async_checkpointer.py:561] Finished blocking save. Time taken: 2.010998s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/0. I0423 03:34:28.192143 140238750848832 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:34:28.192398 140206552385088 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:34:28.192502 140238750848832 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_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776915266.1807418, 'wait_for_prev_duration_secs': 4.124641418457031e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776915266.180934, 'checkpointer_blocking_duration_secs': 2.0111382007598877, 'get_old_steps_start_time': 1776915268.1920886, 'get_old_steps_duration_secs': 2.193450927734375e-05, 'checkpoint_manager_blocking_start_time': 1776915266.1806529, 'checkpoint_manager_blocking_duration_secs': 2.0118227005004883} I0423 03:34:28.192606 140238750848832 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 03:34:28.192653 140238750848832 max_utils.py:750] Memstats: After params initialized: I0423 03:34:28.192700 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 03:34:28.192727 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 03:34:28.192747 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 03:34:28.192765 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 03:34:28.192783 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 03:34:28.192800 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 03:34:28.192817 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 03:34:28.192832 140238750848832 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 03:34:28.564242 140238750848832 metric_logger.py:196] completed step: 0, seconds: 2.218, TFLOP/s/device: 6.124, Tokens/s/device: 923.147, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 03:34:28.565167 140238750848832 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_13_scan_layers_true/tensorboard/' I0423 03:34:28.680365 140238750848832 metric_logger.py:196] completed step: 1, seconds: 2.500, TFLOP/s/device: 5.436, Tokens/s/device: 819.328, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 03:34:28.799597 140238750848832 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 1939.079, Tokens/s/device: 292279.149, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0423 03:34:28.918354 140238750848832 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.326, Tokens/s/device: 17533.946, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0423 03:34:28.973080 140207361885760 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_13_scan_layers_true/checkpoints/0/items/array_metadatas/process_0 I0423 03:34:50.225465 140238750848832 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.743, Tokens/s/device: 17144.531, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0423 03:34:50.243724 140238750848832 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.163, Tokens/s/device: 17207.915, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0423 03:34:50.361085 140238750848832 metric_logger.py:196] completed step: 6, seconds: 21.318, TFLOP/s/device: 0.637, Tokens/s/device: 96.069, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0423 03:34:50.481334 140238750848832 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2614.921, Tokens/s/device: 394149.346, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0423 03:34:50.602145 140238750848832 metric_logger.py:196] completed step: 8, seconds: 0.120, TFLOP/s/device: 113.338, Tokens/s/device: 17083.465, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0423 03:34:50.721685 140238750848832 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 03:34:50.722969 140238750848832 checkpointing.py:776] Waited 0.0015683174133300781 seconds for step 9 to finish before starting checkpointing. I0423 03:34:50.723438 140238750848832 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:35:22.455978 140206753711680 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 54.264778s (commit=53.779849s, array_metadata_write=0.484930s) I0423 03:35:22.456912 140207542240832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 224.868 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 56 seconds) (per-host) I0423 03:35:22.456975 140207542240832 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 54.264977s. I0423 03:35:22.740015 140207542240832 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915267006655757, '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_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:35:22.972026 140207542240832 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 03:35:23.170788 140207347205696 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915267006655757, '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_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:35:23.378590 140207542240832 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:35:23.378797 140207542240832 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.594801s. use_zarr3=True, 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_13_scan_layers_true/checkpoints/0/items I0423 03:35:23.386131 140207542240832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/0/items I0423 03:35:23.639601 140207542240832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/0 I0423 03:35:24.408356 140207542240832 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_13_scan_layers_true/checkpoints/0`. I0423 03:35:24.409003 140207542240832 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 58.228054s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/0 I0423 03:35:24.409104 140207542240832 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 56.217109s. I0423 03:35:24.409304 140206552385088 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:35:24.409413 140206552385088 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:35:24.409480 140206552385088 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:35:24.409531 140206552385088 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 03:35:24.409667 140238750848832 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 03:35:24.409800 140238750848832 checkpoint_manager.py:1441] Waiting for previous save to complete took 33.686405 seconds. If this number is high, consider checkpointing less frequently. I0423 03:35:24.410930 140238750848832 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 03:35:24.411253 140238750848832 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_13_scan_layers_true/checkpoints/9. I0423 03:35:24.577298 140238750848832 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:35:24.577432 140207542240832 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_13_scan_layers_true/checkpoints/9 I0423 03:35:24.577510 140238750848832 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 03:35:25.314023 140207361885760 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_13_scan_layers_true/checkpoints/9/items I0423 03:35:25.737312 140238750848832 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.160554s I0423 03:35:25.737745 140238750848832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 9.900 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 03:35:25.737800 140238750848832 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.246568s (batch_requests_ready=0.082625s, total_serialization_initiated=1.163608s, others=0.000335s) I0423 03:35:25.737861 140238750848832 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.247049s (all_items=0.000025s, per_item={'items': '0.00002503'}, temp_paths=1.247024) I0423 03:35:25.738850 140207171044928 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:35:25.738945 140238750848832 async_checkpointer.py:561] Finished blocking save. Time taken: 1.327971s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/9. I0423 03:35:25.739162 140238750848832 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:35:25.739331 140206552385088 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:35:25.739438 140238750848832 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_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776915290.7233567, 'wait_for_prev_duration_secs': 33.68640470504761, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776915324.4109533, 'checkpointer_blocking_duration_secs': 1.328110694885254, 'get_old_steps_start_time': 1776915325.7390804, 'get_old_steps_duration_secs': 4.38690185546875e-05, 'checkpoint_manager_blocking_start_time': 1776915290.7232747, 'checkpoint_manager_blocking_duration_secs': 35.01614022254944} I0423 03:35:25.739552 140238750848832 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 03:35:25.739609 140238750848832 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:35:26.564472 140207181530688 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_13_scan_layers_true/checkpoints/9/items/array_metadatas/process_0 I0423 03:36:12.105746 140206753711680 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 46.309513s (commit=45.850062s, array_metadata_write=0.459451s) I0423 03:36:12.111125 140207171044928 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 265.368 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host) I0423 03:36:12.111238 140207171044928 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.372247s. I0423 03:36:12.587422 140207171044928 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 03:36:13.029991 140207171044928 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:36:13.030182 140207171044928 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.612910s. use_zarr3=True, 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_13_scan_layers_true/checkpoints/9/items I0423 03:36:13.030833 140207171044928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/9/items I0423 03:36:13.293530 140207171044928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/9 I0423 03:36:14.013098 140207171044928 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_13_scan_layers_true/checkpoints/9`. I0423 03:36:14.013722 140207171044928 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.602755s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_true/checkpoints/9 I0423 03:36:14.013804 140207171044928 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 48.274815s. I0423 03:36:14.014020 140206552385088 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:36:14.014138 140206552385088 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:36:14.014198 140206552385088 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:36:14.014240 140206552385088 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 03:36:14.014399 140238750848832 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 03:36:14.014526 140238750848832 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:36:14.015566 140238750848832 metric_logger.py:196] completed step: 9, seconds: 0.122, TFLOP/s/device: 111.757, Tokens/s/device: 16845.291, total_weights: 16384, loss: 5.915, lm_loss: 5.915, perplexity: 370.653 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops