2026-04-23 03:21:46.216125: 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 00:00:1776914506.230007 2373760 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776914506.234004 2373760 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776914506.246558 2373760 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914506.246574 2373760 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914506.246576 2373760 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914506.246578 2373760 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 03:21:49.916778: 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 03:21:50.051435 139306440165184 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 03:22:56.052803 139306440165184 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 03:22:56.053070 139306440165184 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 03:22:56.053119 139306440165184 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 03:22:56.053152 139306440165184 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 03:22:56.053177 139306440165184 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 03:22:56.053337 139306440165184 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 03:22:56.719143 139306440165184 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 03:22:56.719402 139306440165184 checkpointing.py:677] Setting up checkpoint logger... I0423 03:22:56.719449 139306440165184 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 03:22:56.719484 139306440165184 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 03:22:56.719916 139306440165184 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 0x7eb2855b6fc0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 03:22:59.204730 139306440165184 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 03:22:59.205103 139306440165184 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 0x7eb129d2f1d0>}, handler_registry=None I0423 03:22:59.205322 139306440165184 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eb129d2f1d0>` 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 03:22:59.205362 139306440165184 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb0b8bd3230>` 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 03:22:59.205392 139306440165184 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 0x7eb129d2f1d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eb129d2f1d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb0b8bd3230>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb0b8bd3230>}). I0423 03:22:59.205670 139306440165184 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 03:22:59.205718 139306440165184 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7eb0af704540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 03:22:59.334704 139306440165184 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_11_optimizer_offload_false/checkpoints I0423 03:22:59.334921 139306440165184 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_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7eb0b8bd1310> I0423 03:22:59.335206 139306440165184 checkpointing.py:301] Checkpoint manager created! I0423 03:22:59.724382 139306440165184 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 03:22:59.724468 139306440165184 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:23:00.004801 139306440165184 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:23:00.004881 139306440165184 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:23:00.017167 139306440165184 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:23:00.017215 139306440165184 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:23:00.036349 139306440165184 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 03:23:00.036404 139306440165184 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:23:00.048764 139306440165184 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 03:23:00.048815 139306440165184 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:23:00.061035 139306440165184 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 03:23:00.061096 139306440165184 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:23:00.080491 139306440165184 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 03:23:00.080551 139306440165184 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:23:00.095896 139306440165184 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 03:23:00.095947 139306440165184 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 03:23:00.246324 139306440165184 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 03:23:00.246416 139306440165184 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 03:23:00.872010 139306440165184 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:23:00.872136 139306440165184 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 03:23:00.872181 139306440165184 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 03:23:00.872229 139306440165184 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 03:23:00.872275 139306440165184 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 03:23:00.872304 139306440165184 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 03:23:00.872344 139306440165184 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 03:23:00.872385 139306440165184 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 03:23:00.872416 139306440165184 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 03:23:00.872443 139306440165184 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 03:23:00.872478 139306440165184 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 03:23:00.872515 139306440165184 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 03:23:01.278357 139306440165184 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:23:01.278437 139306440165184 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:23:01.290178 139306440165184 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:23:01.290226 139306440165184 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:23:02.043350 139306440165184 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 03:23:02.043909 139306440165184 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 03:23:02.044070 139306440165184 metric_logger.py:301] number parameters: 1.104 billion I0423 03:23:04.793385 139306440165184 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 03:23:04.911869 139306440165184 checkpointing.py:776] Waited 0.11847209930419922 seconds for step 0 to finish before starting checkpointing. I0423 03:23:04.912477 139306440165184 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:23:04.912635 139306440165184 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 03:23:04.912999 139306440165184 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_11_optimizer_offload_false/checkpoints/0. I0423 03:23:04.995630 139306440165184 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 03:23:05.091528 139306440165184 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:23:05.091615 139306440165184 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 03:23:05.110806 139275173627456 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_11_optimizer_offload_false/checkpoints/0 I0423 03:23:05.850572 139275163141696 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_11_optimizer_offload_false/checkpoints/0/items I0423 03:23:06.005172 139275131684416 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914585760401449, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776914585760401449, "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_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:23:06.330846 2375856 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 03:23:06.330898 2375856 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 03:23:08.156994 139306440165184 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.065826s I0423 03:23:08.162065 139306440165184 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.899 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0423 03:23:08.162127 139306440165184 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.165170s (batch_requests_ready=0.090455s, total_serialization_initiated=3.069757s, others=0.004958s) I0423 03:23:08.162199 139306440165184 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.166095s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=3.166076) I0423 03:23:08.163413 139275184113216 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:23:08.163494 139306440165184 async_checkpointer.py:561] Finished blocking save. Time taken: 3.250819s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/0. I0423 03:23:08.163681 139306440165184 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:23:08.163882 139274533996096 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:23:08.164002 139306440165184 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914584.9124634, 'wait_for_prev_duration_secs': 4.291534423828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914584.9126563, 'checkpointer_blocking_duration_secs': 3.2509520053863525, 'get_old_steps_start_time': 1776914588.1636257, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1776914584.9123845, 'checkpoint_manager_blocking_duration_secs': 3.251589775085449} I0423 03:23:08.164096 139306440165184 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 03:23:08.164140 139306440165184 max_utils.py:750] Memstats: After params initialized: I0423 03:23:08.164189 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 03:23:08.164214 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 03:23:08.164233 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 03:23:08.164251 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 03:23:08.164268 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 03:23:08.164285 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 03:23:08.164301 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 03:23:08.164317 139306440165184 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 03:23:08.565835 139306440165184 metric_logger.py:196] completed step: 0, seconds: 2.749, TFLOP/s/device: 4.943, Tokens/s/device: 745.048, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 03:23:08.566897 139306440165184 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_11_optimizer_offload_false/tensorboard/' I0423 03:23:08.681853 139306440165184 metric_logger.py:196] completed step: 1, seconds: 3.770, TFLOP/s/device: 3.604, Tokens/s/device: 543.211, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0423 03:23:08.801194 139306440165184 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 1829.177, Tokens/s/device: 275713.516, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0423 03:23:08.911691 139275142170176 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_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0423 03:23:08.920570 139306440165184 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.514, Tokens/s/device: 17562.214, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0423 03:23:26.392066 139306440165184 metric_logger.py:196] completed step: 4, seconds: 0.120, TFLOP/s/device: 113.636, Tokens/s/device: 17128.472, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0423 03:23:26.400485 139306440165184 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.094, Tokens/s/device: 17197.511, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0423 03:23:26.517748 139306440165184 metric_logger.py:196] completed step: 6, seconds: 17.473, TFLOP/s/device: 0.778, Tokens/s/device: 117.209, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0423 03:23:26.636656 139306440165184 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2732.729, Tokens/s/device: 411906.677, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0423 03:23:26.755243 139306440165184 metric_logger.py:196] completed step: 8, seconds: 0.118, TFLOP/s/device: 114.752, Tokens/s/device: 17296.713, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0423 03:23:26.872958 139306440165184 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 03:23:26.873977 139306440165184 checkpointing.py:776] Waited 0.0010955333709716797 seconds for step 9 to finish before starting checkpointing. I0423 03:23:26.874277 139306440165184 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:23:57.529670 139274735322688 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 49.366901s (commit=48.872075s, array_metadata_write=0.494826s) I0423 03:23:57.530690 139275184113216 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 240.546 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 52 seconds) (per-host) I0423 03:23:57.530740 139275184113216 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.367204s. I0423 03:23:57.798955 139275184113216 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914585760401449, '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_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:23:58.000503 139275184113216 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 03:23:58.213964 139275131684416 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914585760401449, '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_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:23:58.401911 139275184113216 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:23:58.402094 139275184113216 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.565245s. 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_11_optimizer_offload_false/checkpoints/0/items I0423 03:23:58.402735 139275184113216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/0/items I0423 03:23:58.643830 139275184113216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/0 I0423 03:23:59.344658 139275184113216 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_11_optimizer_offload_false/checkpoints/0`. I0423 03:23:59.345302 139275184113216 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 54.432634s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/0 I0423 03:23:59.345378 139275184113216 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.181842s. I0423 03:23:59.345550 139274533996096 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:23:59.345651 139274533996096 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:23:59.345707 139274533996096 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:23:59.345753 139274533996096 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 03:23:59.345892 139306440165184 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 03:23:59.346016 139306440165184 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.471750 seconds. If this number is high, consider checkpointing less frequently. I0423 03:23:59.347142 139306440165184 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 03:23:59.347430 139306440165184 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_11_optimizer_offload_false/checkpoints/9. I0423 03:23:59.519726 139306440165184 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:23:59.519814 139306440165184 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 03:23:59.579884 139275184113216 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_11_optimizer_offload_false/checkpoints/9 I0423 03:24:00.149399 139306440165184 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.630019s I0423 03:24:00.149805 139306440165184 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 17.176 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 718 milliseconds) (per-host) I0423 03:24:00.149856 139306440165184 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.718568s (batch_requests_ready=0.085190s, total_serialization_initiated=0.633064s, others=0.000314s) I0423 03:24:00.149917 139306440165184 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.719088s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.719070) I0423 03:24:00.150683 139274936649280 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:24:00.150774 139306440165184 async_checkpointer.py:561] Finished blocking save. Time taken: 0.803588s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/9. I0423 03:24:00.150954 139306440165184 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:24:00.151151 139274533996096 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:24:00.151263 139306440165184 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914606.8742397, 'wait_for_prev_duration_secs': 32.471749782562256, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914639.347168, 'checkpointer_blocking_duration_secs': 0.8037023544311523, 'get_old_steps_start_time': 1776914640.1508846, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776914606.8741987, 'checkpoint_manager_blocking_duration_secs': 33.277039527893066} I0423 03:24:00.151374 139306440165184 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 03:24:00.151408 139306440165184 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:24:00.280429 139275142170176 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_11_optimizer_offload_false/checkpoints/9/items I0423 03:24:01.469975 139274972300864 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_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0423 03:24:42.882667 139274961815104 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.147041s (commit=41.704148s, array_metadata_write=0.442894s) I0423 03:24:42.883577 139274936649280 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 290.820 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 43 seconds) (per-host) I0423 03:24:42.883677 139274936649280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.732856s. I0423 03:24:43.386651 139274936649280 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 03:24:43.773523 139274936649280 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:24:43.773730 139274936649280 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.561543s. 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_11_optimizer_offload_false/checkpoints/9/items I0423 03:24:43.774388 139274936649280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/9/items I0423 03:24:44.015789 139274936649280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/9 I0423 03:24:44.847342 139274936649280 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_11_optimizer_offload_false/checkpoints/9`. I0423 03:24:44.847943 139274936649280 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.500763s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_11_optimizer_offload_false/checkpoints/9 I0423 03:24:44.848019 139274936649280 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 44.697199s. I0423 03:24:44.848211 139274533996096 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:24:44.848315 139274533996096 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:24:44.848372 139274533996096 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:24:44.848415 139274533996096 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 03:24:44.848551 139306440165184 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 03:24:44.848675 139306440165184 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:24:44.849514 139306440165184 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.002, Tokens/s/device: 17183.659, 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