2026-04-23 03:18:20.802585: 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 00:00:1776914300.816466 2370660 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776914300.820516 2370660 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776914300.833327 2370660 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914300.833342 2370660 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914300.833344 2370660 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914300.833346 2370660 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 03:18:24.615615: 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 03:18:24.749730 127762536027968 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 03:19:31.731933 127762536027968 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 03:19:31.732187 127762536027968 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 03:19:31.732241 127762536027968 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 03:19:31.732271 127762536027968 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 03:19:31.732443 127762536027968 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 03:19:32.389983 127762536027968 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 03:19:32.390262 127762536027968 checkpointing.py:677] Setting up checkpoint logger... I0423 03:19:32.390307 127762536027968 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 03:19:32.390343 127762536027968 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 03:19:32.390773 127762536027968 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 0x7432c8d4b3b0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 03:19:34.790195 127762536027968 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 03:19:34.790576 127762536027968 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 0x743161b1eea0>}, handler_registry=None I0423 03:19:34.790797 127762536027968 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x743161b1eea0>` 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 03:19:34.790839 127762536027968 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7430e813aea0>` 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 03:19:34.790868 127762536027968 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 0x743161b1eea0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x743161b1eea0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7430e813aea0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7430e813aea0>}). I0423 03:19:34.791157 127762536027968 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 03:19:34.791208 127762536027968 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7430e8170540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 03:19:34.938864 127762536027968 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_10_shardy_false/checkpoints I0423 03:19:34.939212 127762536027968 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_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7430e8139430> I0423 03:19:34.939499 127762536027968 checkpointing.py:301] Checkpoint manager created! I0423 03:19:35.018772 127762536027968 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 03:19:35.021867 127762536027968 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0423 03:19:35.089740 127762536027968 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 03:19:35.115451 127762536027968 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 03:19:35.115516 127762536027968 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 03:19:35.729690 127762536027968 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 03:19:35.729783 127762536027968 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:19:36.028480 127762536027968 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:19:36.028559 127762536027968 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:19:36.041196 127762536027968 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:19:36.041246 127762536027968 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:19:36.060175 127762536027968 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 03:19:36.060230 127762536027968 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:19:36.072980 127762536027968 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 03:19:36.073029 127762536027968 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:19:36.085644 127762536027968 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 03:19:36.085692 127762536027968 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:19:36.105772 127762536027968 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 03:19:36.105830 127762536027968 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:19:36.121779 127762536027968 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 03:19:36.121832 127762536027968 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 03:19:36.275768 127762536027968 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 03:19:36.275867 127762536027968 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 03:19:36.853810 127762536027968 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:19:36.853930 127762536027968 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 03:19:36.853975 127762536027968 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 03:19:36.854023 127762536027968 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 03:19:36.854082 127762536027968 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 03:19:36.854116 127762536027968 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 03:19:36.854157 127762536027968 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 03:19:36.854197 127762536027968 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 03:19:36.854227 127762536027968 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 03:19:36.854254 127762536027968 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 03:19:36.854290 127762536027968 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 03:19:36.854325 127762536027968 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 03:19:37.552259 127762536027968 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:19:37.552342 127762536027968 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:19:37.564337 127762536027968 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:19:37.564388 127762536027968 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:19:38.325351 127762536027968 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 03:19:38.325873 127762536027968 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 03:19:38.326026 127762536027968 metric_logger.py:301] number parameters: 1.104 billion I0423 03:19:42.308761 127762536027968 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 03:19:42.430284 127762536027968 checkpointing.py:776] Waited 0.12151098251342773 seconds for step 0 to finish before starting checkpointing. I0423 03:19:42.430950 127762536027968 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:19:42.431111 127762536027968 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 03:19:42.431466 127762536027968 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_10_shardy_false/checkpoints/0. I0423 03:19:42.526914 127762536027968 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 03:19:42.607084 127637758412352 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_10_shardy_false/checkpoints/0 I0423 03:19:42.620974 127762536027968 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:19:42.621070 127762536027968 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 03:19:43.361595 127637747926592 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_10_shardy_false/checkpoints/0/items I0423 03:19:43.512877 127637716469312 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914383263481638, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776914383263481638, "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_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:19:43.848768 2373182 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 03:19:43.848828 2373182 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 03:19:44.875076 127762536027968 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.254506s I0423 03:19:44.881077 127762536027968 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.245 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 03:19:44.881139 127762536027968 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.353010s (batch_requests_ready=0.089079s, total_serialization_initiated=2.258030s, others=0.005901s) I0423 03:19:44.881222 127762536027968 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.353843s (all_items=0.000031s, per_item={'items': '0.00003123'}, temp_paths=2.353812) I0423 03:19:44.882488 127637768898112 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:19:44.882569 127762536027968 async_checkpointer.py:561] Finished blocking save. Time taken: 2.451417s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/0. I0423 03:19:44.882746 127762536027968 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:19:44.882966 127637118780992 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:19:44.883071 127762536027968 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_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914382.4309375, 'wait_for_prev_duration_secs': 4.124641418457031e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914382.4311335, 'checkpointer_blocking_duration_secs': 2.451540946960449, 'get_old_steps_start_time': 1776914384.8826907, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776914382.4308558, 'checkpoint_manager_blocking_duration_secs': 2.4521772861480713} I0423 03:19:44.883184 127762536027968 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 03:19:44.883225 127762536027968 max_utils.py:750] Memstats: After params initialized: I0423 03:19:44.883274 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 03:19:44.883299 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 03:19:44.883319 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 03:19:44.883336 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 03:19:44.883353 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 03:19:44.883369 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 03:19:44.883387 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 03:19:44.883403 127762536027968 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 03:19:45.253385 127762536027968 metric_logger.py:196] completed step: 0, seconds: 3.982, TFLOP/s/device: 3.412, Tokens/s/device: 514.287, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0423 03:19:45.254294 127762536027968 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_10_shardy_false/tensorboard/' I0423 03:19:45.369981 127762536027968 metric_logger.py:196] completed step: 1, seconds: 2.943, TFLOP/s/device: 4.618, Tokens/s/device: 696.003, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0423 03:19:45.487798 127762536027968 metric_logger.py:196] completed step: 2, seconds: 0.010, TFLOP/s/device: 1370.085, Tokens/s/device: 206514.067, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0423 03:19:45.605319 127762536027968 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.392, Tokens/s/device: 17393.077, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0423 03:19:45.640419 127637726955072 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_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0423 03:20:02.922177 127762536027968 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.168, Tokens/s/device: 17359.316, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0423 03:20:02.940890 127762536027968 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.597, Tokens/s/device: 17424.004, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0423 03:20:03.060477 127762536027968 metric_logger.py:196] completed step: 6, seconds: 17.325, TFLOP/s/device: 0.784, Tokens/s/device: 118.211, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0423 03:20:03.177157 127762536027968 metric_logger.py:196] completed step: 7, seconds: 0.008, TFLOP/s/device: 1616.938, Tokens/s/device: 243722.480, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0423 03:20:03.294001 127762536027968 metric_logger.py:196] completed step: 8, seconds: 0.122, TFLOP/s/device: 111.733, Tokens/s/device: 16841.551, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0423 03:20:03.411977 127762536027968 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 03:20:03.413235 127762536027968 checkpointing.py:776] Waited 0.0015988349914550781 seconds for step 9 to finish before starting checkpointing. I0423 03:20:03.413651 127762536027968 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:20:38.008918 127637387216448 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 53.127079s (commit=52.657109s, array_metadata_write=0.469970s) I0423 03:20:38.009995 127637768898112 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 227.764 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 55 seconds) (per-host) I0423 03:20:38.010079 127637768898112 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 53.127463s. I0423 03:20:38.275610 127637768898112 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914383263481638, '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_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:20:38.485834 127637768898112 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 03:20:38.744332 127637716469312 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914383263481638, '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_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:20:38.884452 127637768898112 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:20:38.889646 127637768898112 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.574105s. 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_10_shardy_false/checkpoints/0/items I0423 03:20:38.890287 127637768898112 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/0/items I0423 03:20:39.130902 127637768898112 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/0 I0423 03:20:39.850164 127637768898112 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_10_shardy_false/checkpoints/0`. I0423 03:20:39.850867 127637768898112 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 57.419719s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/0 I0423 03:20:39.850978 127637768898112 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 54.968364s. I0423 03:20:39.851186 127637118780992 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:20:39.851294 127637118780992 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:20:39.851373 127637118780992 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:20:39.851428 127637118780992 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 03:20:39.851593 127762536027968 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 03:20:39.851720 127762536027968 checkpoint_manager.py:1441] Waiting for previous save to complete took 36.438120 seconds. If this number is high, consider checkpointing less frequently. I0423 03:20:39.853320 127762536027968 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 03:20:39.853634 127762536027968 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_10_shardy_false/checkpoints/9. I0423 03:20:40.043518 127762536027968 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:20:40.043607 127762536027968 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 03:20:40.046266 127637768898112 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_10_shardy_false/checkpoints/9 I0423 03:20:40.764345 127637726955072 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_10_shardy_false/checkpoints/9/items I0423 03:20:41.173505 127762536027968 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.130449s I0423 03:20:41.173961 127762536027968 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 10.055 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 03:20:41.174016 127762536027968 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.227377s (batch_requests_ready=0.092658s, total_serialization_initiated=1.134379s, others=0.000341s) I0423 03:20:41.174096 127762536027968 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.227921s (all_items=0.000025s, per_item={'items': '0.00002480'}, temp_paths=1.227896) I0423 03:20:41.174855 127637588543040 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:20:41.174951 127762536027968 async_checkpointer.py:561] Finished blocking save. Time taken: 1.321582s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/9. I0423 03:20:41.175164 127762536027968 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:20:41.175363 127637118780992 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:20:41.175493 127762536027968 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_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914403.4135673, 'wait_for_prev_duration_secs': 36.438119649887085, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914439.8533478, 'checkpointer_blocking_duration_secs': 1.3217182159423828, 'get_old_steps_start_time': 1776914441.1750827, 'get_old_steps_duration_secs': 4.124641418457031e-05, 'checkpoint_manager_blocking_start_time': 1776914403.4134994, 'checkpoint_manager_blocking_duration_secs': 37.76197052001953} I0423 03:20:41.175606 127762536027968 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 03:20:41.175639 127762536027968 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:20:41.991529 127746034370112 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_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0423 03:21:14.483442 127637387216448 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.250454s (commit=32.776816s, array_metadata_write=0.473638s) I0423 03:21:14.484366 127637588543040 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 365.884 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0423 03:21:14.484473 127637588543040 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.309476s. I0423 03:21:14.940629 127637588543040 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 03:21:15.341138 127637588543040 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:21:15.341355 127637588543040 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.569372s. 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_10_shardy_false/checkpoints/9/items I0423 03:21:15.342025 127637588543040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/9/items I0423 03:21:15.584493 127637588543040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/9 I0423 03:21:16.281472 127637588543040 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_10_shardy_false/checkpoints/9`. I0423 03:21:16.282118 127637588543040 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.428754s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_false/checkpoints/9 I0423 03:21:16.282186 127637588543040 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.107190s. I0423 03:21:16.282359 127637118780992 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:21:16.282456 127637118780992 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:21:16.282510 127637118780992 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:21:16.282550 127637118780992 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 03:21:16.282711 127762536027968 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 03:21:16.282843 127762536027968 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:21:16.283637 127762536027968 metric_logger.py:196] completed step: 9, seconds: 0.120, TFLOP/s/device: 113.050, Tokens/s/device: 17040.112, total_weights: 12300, loss: 8.710, lm_loss: 8.710, perplexity: 6065.008 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops