2026-04-17 13:08:43.040251: 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) I0417 13:08:43.156599 128798702132352 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:09:26.160710 128798702132352 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:09:26.160842 128798702132352 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:09:26.160877 128798702132352 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0417 13:09:26.160902 128798702132352 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:09:26.160986 128798702132352 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:09:26.253271 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:26.654479 128798702132352 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:09:26.654793 128798702132352 checkpointing.py:677] Setting up checkpoint logger... I0417 13:09:26.654846 128798702132352 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:09:26.654894 128798702132352 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:09:26.655510 128798702132352 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 0x7523b1f03b00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:09:29.006298 128798702132352 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:09:29.006477 128798702132352 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 0x751d36fcec30>}, handler_registry=None I0417 13:09:29.006760 128798702132352 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x751d36fcec30>` 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`. I0417 13:09:29.006800 128798702132352 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x751d36130a10>` 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`. I0417 13:09:29.006829 128798702132352 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 0x751d36fcec30>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x751d36fcec30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x751d36130a10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x751d36130a10>}). I0417 13:09:29.007115 128798702132352 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:09:29.007166 128798702132352 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x751d36135e40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:09:29.127309 128798702132352 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints I0417 13:09:29.127536 128798702132352 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_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x751d36131970> I0417 13:09:29.127625 128798702132352 checkpointing.py:301] Checkpoint manager created! I0417 13:09:29.546387 128798702132352 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:09:29.549231 128798702132352 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. I0417 13:09:29.601363 128798702132352 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 I0417 13:09:29.624299 128798702132352 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:09:29.624357 128798702132352 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:09:30.151639 128798702132352 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:09:30.153166 128798702132352 reader.py:262] Creating a tf.data.Dataset reading 2 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0417 13:09:30.169084 128798702132352 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split validation, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:09:30.171633 128798702132352 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:09:30.171681 128798702132352 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:09:30.475915 128798702132352 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:09:30.476015 128798702132352 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:09:30.577247 128798702132352 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:09:30.577323 128798702132352 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:09:30.592306 128798702132352 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:09:30.592355 128798702132352 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:09:30.613838 128798702132352 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:09:30.613895 128798702132352 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:09:30.628887 128798702132352 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:09:30.628937 128798702132352 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:09:30.643821 128798702132352 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:09:30.643869 128798702132352 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:09:30.666892 128798702132352 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:09:30.666948 128798702132352 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:09:30.686000 128798702132352 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:09:30.686073 128798702132352 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:09:30.880031 128798702132352 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:09:30.880134 128798702132352 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 13:09:30.930840 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 I0417 13:09:31.322764 128798702132352 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:09:31.322880 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:09:31.322923 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:09:31.322968 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 13:09:31.323008 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:09:31.323037 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:09:31.323200 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 13:09:31.323242 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0417 13:09:31.323272 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0417 13:09:31.323300 128798702132352 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 13:09:31.323338 128798702132352 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:09:31.323374 128798702132352 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:09:31.762790 128798702132352 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:09:31.762878 128798702132352 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:09:31.777302 128798702132352 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:09:31.777354 128798702132352 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:09:32.049912 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:32.167034 128798702132352 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. I0417 13:09:32.167487 128798702132352 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:09:32.167627 128798702132352 metric_logger.py:301] number parameters: 1.104 billion W0417 13:09:35.544933 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:35.674916 128798702132352 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:09:35.794301 128798702132352 checkpointing.py:776] Waited 0.11937284469604492 seconds for step 0 to finish before starting checkpointing. I0417 13:09:35.794728 128798702132352 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:09:35.794884 128798702132352 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:09:35.795253 128798702132352 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0. I0417 13:09:35.879719 128798702132352 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:09:35.975443 128798702132352 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:09:35.975526 128798702132352 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 I0417 13:09:35.981270 128683654252096 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0 W0417 13:09:36.192577 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.206655 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.217623 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.236641 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.243776 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.251603 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.256811 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.261536 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:36.470957 128798702132352 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.495926s I0417 13:09:36.471333 128798702132352 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 20.894 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 590 milliseconds) (per-host) I0417 13:09:36.471407 128798702132352 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.590760s (batch_requests_ready=0.090937s, total_serialization_initiated=0.499545s, others=0.000278s) I0417 13:09:36.471489 128798702132352 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.591295s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.591276) I0417 13:09:36.472209 128683582948928 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:09:36.472306 128798702132352 async_checkpointer.py:561] Finished blocking save. Time taken: 0.677380s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0. I0417 13:09:36.472514 128798702132352 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:09:36.472682 128683664737856 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:09:36.472786 128798702132352 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776431375.7947147, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776431375.7949078, 'checkpointer_blocking_duration_secs': 0.6775391101837158, 'get_old_steps_start_time': 1776431376.4724631, 'get_old_steps_duration_secs': 1.811981201171875e-05, 'checkpoint_manager_blocking_start_time': 1776431375.7946346, 'checkpoint_manager_blocking_duration_secs': 0.67812180519104} I0417 13:09:36.472901 128798702132352 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:09:36.472938 128798702132352 max_utils.py:750] Memstats: After params initialized: I0417 13:09:36.473009 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:09:36.473032 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:09:36.473064 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:09:36.473082 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:09:36.473098 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:09:36.473115 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:09:36.473130 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:09:36.473146 128798702132352 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:09:36.478999 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.484913 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:09:36.489615 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:36.685571 128683643766336 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/items I0417 13:09:36.816600 128798702132352 metric_logger.py:196] completed step: 0, seconds: 3.507, TFLOP/s/device: 3.874, Tokens/s/device: 584.004, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0417 13:09:36.817487 128798702132352 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/tensorboard/' I0417 13:09:36.849172 128683564074560 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431376598915958, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776431376598915958, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0417 13:09:36.932657 128798702132352 metric_logger.py:196] completed step: 1, seconds: 1.140, TFLOP/s/device: 11.923, Tokens/s/device: 1797.152, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0417 13:09:37.049704 128798702132352 metric_logger.py:196] completed step: 2, seconds: 0.009, TFLOP/s/device: 1441.146, Tokens/s/device: 217225.286, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0417 13:09:37.178266 437291 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:09:37.178337 437291 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:09:38.058096 128683603920448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/items/array_metadatas/process_0 W0417 13:09:43.700743 434038 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:09:43.791710 128798702132352 train.py:588] Completed eval step 0 I0417 13:09:43.833617 128798702132352 train.py:588] Completed eval step 1 I0417 13:09:43.838302 128798702132352 metric_logger.py:216] eval metrics after step: 4, loss=8.979, perplexity=7936.996, total_weights=26703.0 I0417 13:09:43.838809 128798702132352 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.667, Tokens/s/device: 17434.535, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0417 13:10:03.261294 128798702132352 metric_logger.py:196] completed step: 4, seconds: 0.117, TFLOP/s/device: 115.967, Tokens/s/device: 17479.772, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0417 13:10:03.272251 128798702132352 metric_logger.py:196] completed step: 5, seconds: 6.788, TFLOP/s/device: 2.002, Tokens/s/device: 301.721, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0417 13:10:03.388559 128798702132352 metric_logger.py:196] completed step: 6, seconds: 19.424, TFLOP/s/device: 0.699, Tokens/s/device: 105.434, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0417 13:10:03.505769 128798702132352 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 2106.858, Tokens/s/device: 317568.615, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0417 13:10:04.194459 128798702132352 train.py:588] Completed eval step 0 I0417 13:10:04.234437 128798702132352 train.py:588] Completed eval step 1 I0417 13:10:04.237539 128798702132352 metric_logger.py:216] eval metrics after step: 9, loss=8.653, perplexity=5725.923, total_weights=26703.0 I0417 13:10:04.237986 128798702132352 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.956, Tokens/s/device: 17478.131, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0417 13:10:04.239738 128798702132352 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:10:04.240348 128798702132352 checkpointing.py:776] Waited 0.0006086826324462891 seconds for step 9 to finish before starting checkpointing. I0417 13:10:04.240633 128798702132352 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:10:05.447706 128683593434688 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 28.267142s (commit=27.798264s, array_metadata_write=0.468878s) I0417 13:10:05.448841 128683582948928 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 427.379 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 29 seconds) (per-host) I0417 13:10:05.448964 128683582948928 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.976571s. I0417 13:10:05.709005 128683582948928 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431376598915958, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0417 13:10:05.884692 128683582948928 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:10:06.081806 128683564074560 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431376598915958, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0417 13:10:06.285533 128683582948928 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:10:06.285714 128683582948928 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.537939s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/items I0417 13:10:06.286862 128683582948928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0/items I0417 13:10:06.532861 128683582948928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0 I0417 13:10:07.213276 128683582948928 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0`. I0417 13:10:07.213892 128683582948928 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 31.418969s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/0 I0417 13:10:07.213985 128683582948928 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 30.741598s. I0417 13:10:07.214199 128683664737856 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:10:07.214307 128683664737856 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:10:07.214377 128683664737856 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:10:07.214437 128683664737856 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:10:07.214561 128798702132352 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. W0417 13:10:07.214698 128798702132352 checkpoint_manager.py:1441] Waiting for previous save to complete took 2.974092 seconds. If this number is high, consider checkpointing less frequently. I0417 13:10:07.215770 128798702132352 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:10:07.216093 128798702132352 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9. I0417 13:10:07.382192 128683664737856 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9 I0417 13:10:07.395105 128798702132352 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:10:07.395202 128798702132352 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 I0417 13:10:08.069363 128683593434688 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9/items I0417 13:10:10.473341 128798702132352 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.078632s I0417 13:10:10.477609 128798702132352 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.887 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0417 13:10:10.477673 128798702132352 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.175220s (batch_requests_ready=0.088161s, total_serialization_initiated=3.082899s, others=0.004161s) I0417 13:10:10.477746 128798702132352 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.176004s (all_items=0.000023s, per_item={'items': '0.00002337'}, temp_paths=3.175981) I0417 13:10:10.479082 128683664737856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:10:10.479218 128798702132352 async_checkpointer.py:561] Finished blocking save. Time taken: 3.263400s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9. I0417 13:10:10.479434 128798702132352 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:10:10.479681 128683582948928 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:10:10.479799 128798702132352 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776431404.2405734, 'wait_for_prev_duration_secs': 2.9740920066833496, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776431407.2157943, 'checkpointer_blocking_duration_secs': 3.2635393142700195, 'get_old_steps_start_time': 1776431410.4793525, 'get_old_steps_duration_secs': 4.076957702636719e-05, 'checkpoint_manager_blocking_start_time': 1776431404.2405086, 'checkpoint_manager_blocking_duration_secs': 6.239264965057373} I0417 13:10:10.479913 128798702132352 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:10:10.479946 128798702132352 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:10:11.260803 128683603920448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9/items/array_metadatas/process_0 I0417 13:10:43.661924 128683593434688 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.183577s (commit=32.723830s, array_metadata_write=0.459747s) I0417 13:10:43.663030 128683664737856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 347.541 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0417 13:10:43.663148 128683664737856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.183882s. I0417 13:10:44.083265 128683664737856 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:10:44.514489 128683664737856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:10:44.514669 128683664737856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570946s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9/items I0417 13:10:44.515373 128683664737856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9/items I0417 13:10:44.761573 128683664737856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9 I0417 13:10:45.432801 128683664737856 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9`. I0417 13:10:45.433446 128683664737856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.217632s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_07_eval/checkpoints/9 I0417 13:10:45.433523 128683664737856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.954262s. I0417 13:10:45.433711 128683582948928 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:10:45.433820 128683582948928 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:10:45.433896 128683582948928 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:10:45.433947 128683582948928 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:10:45.434194 128798702132352 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. I0417 13:10:45.434559 128798702132352 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:10:45.435416 128798702132352 metric_logger.py:196] completed step: 9, seconds: 0.117, TFLOP/s/device: 115.915, Tokens/s/device: 17472.017, 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