2026-04-23 02:45:34.148569: 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 00:00:1776912334.162249 2338298 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776912334.166264 2338298 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776912334.178722 2338298 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776912334.178737 2338298 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776912334.178738 2338298 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776912334.178740 2338298 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 02:45:37.856774: 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 02:45:37.991561 130263227504448 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 02:46:38.611407 130263227504448 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 02:46:38.611636 130263227504448 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 02:46:38.611684 130263227504448 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 02:46:38.611713 130263227504448 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 02:46:38.611738 130263227504448 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 02:46:38.611893 130263227504448 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 02:46:39.278220 130263227504448 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 02:46:39.278481 130263227504448 checkpointing.py:677] Setting up checkpoint logger... I0423 02:46:39.278526 130263227504448 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 02:46:39.278562 130263227504448 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 02:46:39.278992 130263227504448 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 0x7678fbda9b80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 02:46:41.701225 130263227504448 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 02:46:41.701580 130263227504448 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 0x767728f167e0>}, handler_registry=None I0423 02:46:41.701804 130263227504448 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x767728f167e0>` 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 02:46:41.701845 130263227504448 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x767728f15880>` 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 02:46:41.701875 130263227504448 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 0x767728f167e0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x767728f167e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x767728f15880>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x767728f15880>}). I0423 02:46:41.702177 130263227504448 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 02:46:41.702227 130263227504448 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x767728f4c540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 02:46:41.854250 130263227504448 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_02_synthetic/checkpoints I0423 02:46:41.854464 130263227504448 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_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x767728f15ca0> I0423 02:46:41.854738 130263227504448 checkpointing.py:301] Checkpoint manager created! I0423 02:46:42.242397 130263227504448 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 02:46:42.242487 130263227504448 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:46:42.520879 130263227504448 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:46:42.520967 130263227504448 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:46:42.533216 130263227504448 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:46:42.533265 130263227504448 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:46:42.552084 130263227504448 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 02:46:42.552140 130263227504448 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:46:42.564383 130263227504448 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 02:46:42.564433 130263227504448 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:46:42.576604 130263227504448 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 02:46:42.576654 130263227504448 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:46:42.596000 130263227504448 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 02:46:42.596067 130263227504448 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:46:42.611583 130263227504448 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 02:46:42.611632 130263227504448 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 02:46:42.760954 130263227504448 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 02:46:42.761053 130263227504448 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 02:46:43.401330 130263227504448 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:46:43.401445 130263227504448 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 02:46:43.401489 130263227504448 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 02:46:43.401535 130263227504448 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 02:46:43.401578 130263227504448 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 02:46:43.401607 130263227504448 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 02:46:43.401652 130263227504448 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 02:46:43.401695 130263227504448 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 02:46:43.401725 130263227504448 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 02:46:43.401753 130263227504448 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 02:46:43.401788 130263227504448 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 02:46:43.401823 130263227504448 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 02:46:43.807119 130263227504448 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:46:43.807201 130263227504448 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:46:43.818798 130263227504448 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:46:43.818849 130263227504448 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:46:44.576220 130263227504448 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 02:46:44.576777 130263227504448 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 02:46:44.576926 130263227504448 metric_logger.py:301] number parameters: 1.104 billion I0423 02:46:47.339848 130263227504448 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 02:46:47.457767 130263227504448 checkpointing.py:776] Waited 0.11791086196899414 seconds for step 0 to finish before starting checkpointing. I0423 02:46:47.458367 130263227504448 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:46:47.458514 130263227504448 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 02:46:47.458895 130263227504448 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_02_synthetic/checkpoints/0. I0423 02:46:47.555473 130263227504448 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 02:46:47.641580 130231851550272 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_02_synthetic/checkpoints/0 I0423 02:46:47.653022 130263227504448 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 02:46:47.653121 130263227504448 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 02:46:48.391954 130231841064512 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_02_synthetic/checkpoints/0/items I0423 02:46:48.547407 130231809607232 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776912408306346302, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776912408306346302, "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_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0423 02:46:48.881186 2340378 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 02:46:48.881238 2340378 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 02:46:49.586480 130263227504448 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.933855s I0423 02:46:49.591167 130263227504448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.065 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 02:46:49.591230 130263227504448 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.034830s (batch_requests_ready=0.092459s, total_serialization_initiated=1.937783s, others=0.004588s) I0423 02:46:49.591301 130263227504448 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.035372s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=2.035352) I0423 02:46:49.592749 130231862036032 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:46:49.592851 130263227504448 async_checkpointer.py:561] Finished blocking save. Time taken: 2.134293s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/0. I0423 02:46:49.593029 130263227504448 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:46:49.593212 130231021078080 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:46:49.593318 130263227504448 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_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776912407.4583545, 'wait_for_prev_duration_secs': 4.124641418457031e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776912407.4585352, 'checkpointer_blocking_duration_secs': 2.1344246864318848, 'get_old_steps_start_time': 1776912409.5929756, 'get_old_steps_duration_secs': 2.002716064453125e-05, 'checkpoint_manager_blocking_start_time': 1776912407.4582741, 'checkpoint_manager_blocking_duration_secs': 2.1350183486938477} I0423 02:46:49.593395 130263227504448 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 02:46:49.593435 130263227504448 max_utils.py:750] Memstats: After params initialized: I0423 02:46:49.593482 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 02:46:49.593506 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 02:46:49.593526 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 02:46:49.593545 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 02:46:49.593563 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 02:46:49.593581 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 02:46:49.593598 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 02:46:49.593614 130263227504448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 02:46:50.021287 130263227504448 metric_logger.py:196] completed step: 0, seconds: 2.762, TFLOP/s/device: 4.919, Tokens/s/device: 741.385, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 02:46:50.022130 130263227504448 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_02_synthetic/tensorboard/' I0423 02:46:50.138141 130263227504448 metric_logger.py:196] completed step: 1, seconds: 2.679, TFLOP/s/device: 5.071, Tokens/s/device: 764.364, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 02:46:50.257354 130263227504448 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 1991.372, Tokens/s/device: 300161.219, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0423 02:46:50.350287 130231820092992 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_02_synthetic/checkpoints/0/items/array_metadatas/process_0 I0423 02:46:50.376770 130263227504448 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.199, Tokens/s/device: 17364.026, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0423 02:47:07.223523 130263227504448 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.105, Tokens/s/device: 17199.100, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0423 02:47:07.232171 130263227504448 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 113.756, Tokens/s/device: 17146.541, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0423 02:47:07.349313 130263227504448 metric_logger.py:196] completed step: 6, seconds: 16.848, TFLOP/s/device: 0.806, Tokens/s/device: 121.555, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0423 02:47:07.468014 130263227504448 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2658.409, Tokens/s/device: 400704.363, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0423 02:47:07.586829 130263227504448 metric_logger.py:196] completed step: 8, seconds: 0.118, TFLOP/s/device: 114.977, Tokens/s/device: 17330.524, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0423 02:47:07.704542 130263227504448 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 02:47:07.705491 130263227504448 checkpointing.py:776] Waited 0.0009725093841552734 seconds for step 9 to finish before starting checkpointing. I0423 02:47:07.705811 130263227504448 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:47:13.468690 130231222404672 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 23.876790s (commit=23.402454s, array_metadata_write=0.474336s) I0423 02:47:13.469807 130231862036032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 487.656 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host) I0423 02:47:13.469930 130231862036032 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 23.877028s. I0423 02:47:13.711637 130231862036032 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776912408306346302, '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_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0423 02:47:13.938888 130231862036032 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 02:47:14.135970 130231809607232 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776912408306346302, '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_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0423 02:47:14.326222 130231862036032 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:47:14.326393 130231862036032 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573253s. 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_02_synthetic/checkpoints/0/items I0423 02:47:14.327114 130231862036032 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/0/items I0423 02:47:14.569440 130231862036032 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/0 I0423 02:47:15.255994 130231862036032 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_02_synthetic/checkpoints/0`. I0423 02:47:15.256648 130231862036032 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 27.798096s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/0 I0423 02:47:15.256735 130231862036032 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 25.663838s. I0423 02:47:15.256944 130231021078080 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:47:15.257067 130231021078080 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:47:15.257133 130231021078080 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:47:15.257184 130231021078080 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 02:47:15.257311 130263227504448 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 02:47:15.257425 130263227504448 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.551645 seconds. If this number is high, consider checkpointing less frequently. I0423 02:47:15.258728 130263227504448 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 02:47:15.259025 130263227504448 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_02_synthetic/checkpoints/9. I0423 02:47:15.427908 130231862036032 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_02_synthetic/checkpoints/9 I0423 02:47:15.458730 130263227504448 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 02:47:15.458824 130263227504448 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 02:47:16.118313 130231222404672 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_02_synthetic/checkpoints/9/items I0423 02:47:16.218470 130263227504448 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.760113s I0423 02:47:16.218876 130263227504448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 14.077 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 876 milliseconds) (per-host) I0423 02:47:16.218933 130263227504448 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.876711s (batch_requests_ready=0.113200s, total_serialization_initiated=0.763192s, others=0.000319s) I0423 02:47:16.219001 130263227504448 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.877440s (all_items=0.000029s, per_item={'items': '0.00002909'}, temp_paths=0.877411) I0423 02:47:16.219918 130231639737920 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:47:16.220013 130263227504448 async_checkpointer.py:561] Finished blocking save. Time taken: 0.961240s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/9. I0423 02:47:16.220221 130263227504448 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:47:16.220443 130263227504448 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_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776912427.7057533, 'wait_for_prev_duration_secs': 7.551645040512085, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776912435.2587523, 'checkpointer_blocking_duration_secs': 0.9613792896270752, 'get_old_steps_start_time': 1776912436.220145, 'get_old_steps_duration_secs': 3.910064697265625e-05, 'checkpoint_manager_blocking_start_time': 1776912427.7056966, 'checkpoint_manager_blocking_duration_secs': 8.514726400375366} I0423 02:47:16.220514 130263227504448 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 02:47:16.220547 130263227504448 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:47:16.283040 130231021078080 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:47:17.370228 130231650223680 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_02_synthetic/checkpoints/9/items/array_metadatas/process_0 I0423 02:47:54.496244 130231820092992 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.879829s (commit=37.399297s, array_metadata_write=0.480531s) I0423 02:47:54.497225 130231639737920 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 322.738 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0423 02:47:54.497327 130231639737920 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.277256s. I0423 02:47:54.997492 130231639737920 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 02:47:55.404598 130231639737920 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:47:55.404825 130231639737920 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.578615s. 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_02_synthetic/checkpoints/9/items I0423 02:47:55.405532 130231639737920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/9/items I0423 02:47:55.665370 130231639737920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/9 I0423 02:47:56.392128 130231639737920 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_02_synthetic/checkpoints/9`. I0423 02:47:56.392760 130231639737920 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.133992s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_02_synthetic/checkpoints/9 I0423 02:47:56.392830 130231639737920 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.172760s. I0423 02:47:56.392971 130231021078080 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:47:56.393088 130231021078080 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:47:56.393136 130231021078080 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:47:56.393172 130231021078080 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 02:47:56.393333 130263227504448 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 02:47:56.393458 130263227504448 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:47:56.394360 130263227504448 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.754, Tokens/s/device: 17297.005, 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