2026-04-23 03:15:09.870338: 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 00:00:1776914109.884203 2366819 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776914109.888203 2366819 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776914109.900999 2366819 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914109.901015 2366819 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914109.901017 2366819 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914109.901019 2366819 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 03:15:13.558610: 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 03:15:13.693727 135990608828224 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 03:16:17.247965 135990608828224 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 03:16:17.248220 135990608828224 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 03:16:17.248272 135990608828224 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 03:16:17.248301 135990608828224 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 03:16:17.248475 135990608828224 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 03:16:17.916303 135990608828224 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 03:16:17.916546 135990608828224 checkpointing.py:677] Setting up checkpoint logger... I0423 03:16:17.916590 135990608828224 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 03:16:17.916625 135990608828224 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 03:16:17.917064 135990608828224 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 0x7bae7e5b1a90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 03:16:20.354483 135990608828224 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 03:16:20.354829 135990608828224 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 0x7bacaef0e780>}, handler_registry=None I0423 03:16:20.355069 135990608828224 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bacaef0e780>` 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 03:16:20.355114 135990608828224 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bacaef0d880>` 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 03:16:20.355143 135990608828224 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 0x7bacaef0e780>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bacaef0e780>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bacaef0d880>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bacaef0d880>}). I0423 03:16:20.355417 135990608828224 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 03:16:20.355467 135990608828224 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7bacaef3c540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 03:16:20.474202 135990608828224 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_true/checkpoints I0423 03:16:20.474414 135990608828224 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_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bad22da46b0> I0423 03:16:20.474688 135990608828224 checkpointing.py:301] Checkpoint manager created! I0423 03:16:20.551859 135990608828224 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 03:16:20.554616 135990608828224 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 03:16:20.614004 135990608828224 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 03:16:20.636391 135990608828224 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 03:16:20.636456 135990608828224 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 03:16:21.242551 135990608828224 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 03:16:21.242642 135990608828224 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:16:21.540239 135990608828224 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:16:21.540318 135990608828224 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:16:21.552988 135990608828224 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:16:21.553035 135990608828224 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:16:21.572128 135990608828224 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 03:16:21.572185 135990608828224 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:16:21.584895 135990608828224 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 03:16:21.584946 135990608828224 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:16:21.597637 135990608828224 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 03:16:21.597685 135990608828224 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:16:21.617508 135990608828224 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 03:16:21.617562 135990608828224 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:16:21.633388 135990608828224 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 03:16:21.633441 135990608828224 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 03:16:21.787895 135990608828224 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 03:16:21.787987 135990608828224 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 03:16:22.426210 135990608828224 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:16:22.426322 135990608828224 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 03:16:22.426365 135990608828224 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 03:16:22.426412 135990608828224 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 03:16:22.426454 135990608828224 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 03:16:22.426483 135990608828224 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 03:16:22.426524 135990608828224 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 03:16:22.426564 135990608828224 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 03:16:22.426593 135990608828224 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 03:16:22.426621 135990608828224 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 03:16:22.426656 135990608828224 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 03:16:22.426690 135990608828224 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 03:16:23.108630 135990608828224 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:16:23.108712 135990608828224 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:16:23.121854 135990608828224 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:16:23.121903 135990608828224 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:16:23.873368 135990608828224 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 03:16:23.873881 135990608828224 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 03:16:23.874026 135990608828224 metric_logger.py:301] number parameters: 1.104 billion I0423 03:16:28.556146 135990608828224 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 03:16:28.678119 135990608828224 checkpointing.py:776] Waited 0.12197065353393555 seconds for step 0 to finish before starting checkpointing. I0423 03:16:28.678756 135990608828224 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:16:28.678920 135990608828224 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 03:16:28.679349 135990608828224 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_true/checkpoints/0. I0423 03:16:28.778069 135990608828224 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 03:16:28.866986 135866078987840 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_true/checkpoints/0 I0423 03:16:28.873999 135990608828224 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:16:28.874101 135990608828224 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 03:16:29.612478 135866068502080 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_true/checkpoints/0/items I0423 03:16:29.759086 135865682626112 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914189517835474, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776914189517835474, "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_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:16:30.084348 2370061 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 03:16:30.084402 2370061 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 03:16:31.744563 135990608828224 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.870896s I0423 03:16:31.745169 135990608828224 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.161 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 03:16:31.745224 135990608828224 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.965919s (batch_requests_ready=0.090955s, total_serialization_initiated=2.874463s, others=0.000501s) I0423 03:16:31.745292 135990608828224 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.966766s (all_items=0.000022s, per_item={'items': '0.00002217'}, temp_paths=2.966744) I0423 03:16:31.750525 135865145755200 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:16:31.750672 135990608828224 async_checkpointer.py:561] Finished blocking save. Time taken: 3.071712s. 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_true/checkpoints/0. I0423 03:16:31.750879 135990608828224 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:16:31.751073 135864944428608 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:16:31.751137 135990608828224 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914188.6787422, 'wait_for_prev_duration_secs': 4.38690185546875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914188.6789412, 'checkpointer_blocking_duration_secs': 3.071856737136841, 'get_old_steps_start_time': 1776914191.7508147, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1776914188.6786466, 'checkpoint_manager_blocking_duration_secs': 3.072465419769287} I0423 03:16:31.751249 135990608828224 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 03:16:31.751296 135990608828224 max_utils.py:750] Memstats: After params initialized: I0423 03:16:31.751343 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 03:16:31.751368 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 03:16:31.751388 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 03:16:31.751406 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 03:16:31.751425 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 03:16:31.751442 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 03:16:31.751459 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 03:16:31.751476 135990608828224 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 03:16:32.131086 135990608828224 metric_logger.py:196] completed step: 0, seconds: 4.682, TFLOP/s/device: 2.902, Tokens/s/device: 437.458, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0423 03:16:32.131997 135990608828224 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_true/tensorboard/' I0423 03:16:32.247465 135990608828224 metric_logger.py:196] completed step: 1, seconds: 3.573, TFLOP/s/device: 3.803, Tokens/s/device: 573.251, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0423 03:16:32.365308 135990608828224 metric_logger.py:196] completed step: 2, seconds: 0.010, TFLOP/s/device: 1303.072, Tokens/s/device: 196413.158, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0423 03:16:32.482723 135990608828224 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.464, Tokens/s/device: 17404.014, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0423 03:16:32.570625 135865909118528 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_true/checkpoints/0/items/array_metadatas/process_0 I0423 03:16:49.067969 135990608828224 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.137, Tokens/s/device: 17354.756, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0423 03:16:49.078019 135990608828224 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.556, Tokens/s/device: 17417.928, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0423 03:16:49.195537 135990608828224 metric_logger.py:196] completed step: 6, seconds: 16.584, TFLOP/s/device: 0.819, Tokens/s/device: 123.489, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0423 03:16:49.313646 135990608828224 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2006.369, Tokens/s/device: 302421.737, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0423 03:16:49.431618 135990608828224 metric_logger.py:196] completed step: 8, seconds: 0.119, TFLOP/s/device: 114.329, Tokens/s/device: 17232.965, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0423 03:16:49.548586 135990608828224 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 03:16:49.549439 135990608828224 checkpointing.py:776] Waited 0.0008776187896728516 seconds for step 9 to finish before starting checkpointing. I0423 03:16:49.549729 135990608828224 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:17:09.163432 135865336596032 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.413415s (commit=36.935282s, array_metadata_write=0.478133s) I0423 03:17:09.164423 135865145755200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 312.907 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0423 03:17:09.164526 135865145755200 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.413809s. I0423 03:17:09.457352 135865145755200 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914189517835474, '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_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:17:09.680885 135865145755200 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 03:17:09.898203 135865682626112 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776914189517835474, '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_true/checkpoints/0/_CHECKPOINT_METADATA I0423 03:17:10.088706 135865145755200 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:17:10.088865 135865145755200 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.586981s. 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_true/checkpoints/0/items I0423 03:17:10.089635 135865145755200 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_true/checkpoints/0/items I0423 03:17:10.328825 135865145755200 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_true/checkpoints/0 I0423 03:17:11.032956 135865145755200 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_true/checkpoints/0`. I0423 03:17:11.033657 135865145755200 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.354703s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_true/checkpoints/0 I0423 03:17:11.033731 135865145755200 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.283015s. I0423 03:17:11.033836 135864944428608 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:17:11.033890 135864944428608 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:17:11.033944 135864944428608 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:17:11.033985 135864944428608 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 03:17:11.034155 135990608828224 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 03:17:11.034281 135990608828224 checkpoint_manager.py:1441] Waiting for previous save to complete took 21.484557 seconds. If this number is high, consider checkpointing less frequently. I0423 03:17:11.035343 135990608828224 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 03:17:11.035628 135990608828224 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_true/checkpoints/9. I0423 03:17:11.212742 135865145755200 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_true/checkpoints/9 I0423 03:17:11.216717 135990608828224 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 03:17:11.216807 135990608828224 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 03:17:11.916494 135865336596032 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_true/checkpoints/9/items I0423 03:17:14.902381 135990608828224 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.686065s I0423 03:17:14.906747 135990608828224 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.272 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0423 03:17:14.906816 135990608828224 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.771979s (batch_requests_ready=0.078682s, total_serialization_initiated=3.689027s, others=0.004270s) I0423 03:17:14.906901 135990608828224 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.772731s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=3.772714) I0423 03:17:14.908329 135864944428608 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:17:14.908425 135990608828224 async_checkpointer.py:561] Finished blocking save. Time taken: 3.873036s. 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_true/checkpoints/9. I0423 03:17:14.908621 135990608828224 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:17:14.908802 135865909118528 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:17:14.908903 135990608828224 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776914209.5496974, 'wait_for_prev_duration_secs': 21.484557151794434, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776914231.0353692, 'checkpointer_blocking_duration_secs': 3.873164415359497, 'get_old_steps_start_time': 1776914234.9085503, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776914209.5496504, 'checkpoint_manager_blocking_duration_secs': 25.35922908782959} I0423 03:17:14.909012 135990608828224 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 03:17:14.909059 135990608828224 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:17:15.675845 135973964875328 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_true/checkpoints/9/items/array_metadatas/process_0 I0423 03:17:49.845444 135973954389568 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.938081s (commit=34.456850s, array_metadata_write=0.481231s) I0423 03:17:49.846491 135864944428608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 326.434 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0423 03:17:49.846600 135864944428608 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.938130s. I0423 03:17:50.431277 135864944428608 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 03:17:50.843169 135864944428608 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:17:50.843370 135864944428608 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.585013s. 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_true/checkpoints/9/items I0423 03:17:50.843957 135864944428608 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_true/checkpoints/9/items I0423 03:17:51.088308 135864944428608 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_true/checkpoints/9 I0423 03:17:51.820997 135864944428608 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_true/checkpoints/9`. I0423 03:17:51.821625 135864944428608 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.786244s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_10_shardy_true/checkpoints/9 I0423 03:17:51.821704 135864944428608 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.913236s. I0423 03:17:51.821898 135865909118528 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:17:51.822004 135865909118528 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:17:51.822071 135865909118528 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:17:51.822110 135865909118528 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 03:17:51.822250 135990608828224 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 03:17:51.822385 135990608828224 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:17:51.823193 135990608828224 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.269, Tokens/s/device: 17374.632, 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