2026-04-21 15:08:26.508804: 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 00:00:1776784106.519312 1498035 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776784106.522332 1498035 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776784106.532166 1498035 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784106.532178 1498035 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784106.532179 1498035 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784106.532180 1498035 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 15:08:29.913608: 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) I0421 15:08:30.029737 133805176765568 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:08:56.930334 133805176765568 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:08:56.930542 133805176765568 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:08:56.930591 133805176765568 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 I0421 15:08:56.930619 133805176765568 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:08:56.930641 133805176765568 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 15:08:56.930752 133805176765568 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:08:57.285310 133805176765568 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:08:57.285681 133805176765568 checkpointing.py:677] Setting up checkpoint logger... I0421 15:08:57.285727 133805176765568 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:08:57.285815 133805176765568 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:08:57.286654 133805176765568 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 0x79b1a85a6e40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:08:59.710936 133805176765568 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:08:59.711227 133805176765568 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 0x79afff376360>}, handler_registry=None I0421 15:08:59.711615 133805176765568 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79afff376360>` 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`. I0421 15:08:59.711658 133805176765568 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afff376900>` 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`. I0421 15:08:59.711687 133805176765568 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 0x79afff376360>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79afff376360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afff376900>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afff376900>}). I0421 15:08:59.711973 133805176765568 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:08:59.712055 133805176765568 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79afff3951c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 15:08:59.838585 133805176765568 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints I0421 15:08:59.838851 133805176765568 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=5), 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_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79afff376840> I0421 15:08:59.838943 133805176765568 checkpointing.py:301] Checkpoint manager created! I0421 15:09:00.476216 133805176765568 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:09:00.476328 133805176765568 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:09:00.758251 133805176765568 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:09:00.758337 133805176765568 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:09:00.770323 133805176765568 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:09:00.770374 133805176765568 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:09:00.789319 133805176765568 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:09:00.789377 133805176765568 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:09:00.801385 133805176765568 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:09:00.801442 133805176765568 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:09:00.813282 133805176765568 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:09:00.813330 133805176765568 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:09:00.832437 133805176765568 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:09:00.832498 133805176765568 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:09:00.847627 133805176765568 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:09:00.847680 133805176765568 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:09:00.996123 133805176765568 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:09:00.996229 133805176765568 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 I0421 15:09:01.588244 133805176765568 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:09:01.588355 133805176765568 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:09:01.588397 133805176765568 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:09:01.588440 133805176765568 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 15:09:01.588480 133805176765568 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:09:01.588510 133805176765568 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:09:01.588551 133805176765568 maxtext_utils.py:1654] 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) I0421 15:09:01.588592 133805176765568 maxtext_utils.py:1654] 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') I0421 15:09:01.588621 133805176765568 maxtext_utils.py:1654] 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) I0421 15:09:01.588648 133805176765568 maxtext_utils.py:1654] 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) I0421 15:09:01.588684 133805176765568 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:09:01.588719 133805176765568 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:09:01.992770 133805176765568 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:09:01.992859 133805176765568 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:09:02.004134 133805176765568 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:09:02.004186 133805176765568 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:09:10.054393 133805176765568 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. I0421 15:09:10.055078 133805176765568 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:09:10.055222 133805176765568 metric_logger.py:301] number parameters: 1.104 billion I0421 15:09:19.080997 133805176765568 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0421 15:09:19.207998 133805176765568 checkpointing.py:776] Waited 0.12698149681091309 seconds for step 0 to finish before starting checkpointing. I0421 15:09:19.208500 133805176765568 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:09:19.208648 133805176765568 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0421 15:09:19.209076 133805176765568 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0. I0421 15:09:19.301725 133805176765568 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 15:09:19.385576 133805176765568 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:09:19.385681 133805176765568 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 I0421 15:09:19.399106 133720644257344 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0 I0421 15:09:20.167850 133721038521920 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/items I0421 15:09:20.300484 133805176765568 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.915421s I0421 15:09:20.301079 133805176765568 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.364 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 998 milliseconds) (per-host) I0421 15:09:20.301249 133805176765568 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.998432s (batch_requests_ready=0.078632s, total_serialization_initiated=0.919248s, others=0.000552s) I0421 15:09:20.301377 133805176765568 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.999152s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.999132) I0421 15:09:20.302458 133720981898816 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:09:20.302567 133805176765568 async_checkpointer.py:561] Finished blocking save. Time taken: 1.093879s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0. I0421 15:09:20.302770 133805176765568 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:09:20.302992 133805176765568 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784159.2084873, 'wait_for_prev_duration_secs': 3.886222839355469e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784159.2086687, 'checkpointer_blocking_duration_secs': 1.094017505645752, 'get_old_steps_start_time': 1776784160.3027046, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776784159.2084038, 'checkpoint_manager_blocking_duration_secs': 1.0945634841918945} I0421 15:09:20.303088 133805176765568 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0421 15:09:20.303132 133805176765568 max_utils.py:750] Memstats: After params initialized: I0421 15:09:20.303210 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:09:20.303237 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:09:20.303258 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:09:20.303277 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:09:20.303294 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:09:20.303312 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:09:20.303329 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:09:20.303349 133805176765568 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:09:20.321969 133721002870336 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784160076845627, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776784160076845627, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0421 15:09:20.322185 133720971413056 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:09:20.577923 133805176765568 metric_logger.py:196] completed step: 0, seconds: 9.025, TFLOP/s/device: 1.506, Tokens/s/device: 226.931, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0421 15:09:20.578763 133805176765568 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/tensorboard/' I0421 15:09:20.658715 1500525 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:09:20.658774 1500525 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:09:20.944623 133805176765568 metric_logger.py:196] completed step: 1, seconds: 1.495, TFLOP/s/device: 9.090, Tokens/s/device: 1370.135, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0421 15:09:20.954200 133805176765568 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2186.535, Tokens/s/device: 329578.371, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0421 15:09:21.072148 133805176765568 metric_logger.py:196] completed step: 3, seconds: 0.371, TFLOP/s/device: 36.591, Tokens/s/device: 5515.458, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0421 15:09:21.077306 133805176765568 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0421 15:09:21.309585 133805176765568 checkpointing.py:776] Waited 0.23223567008972168 seconds for step 5 to finish before starting checkpointing. I0421 15:09:21.310010 133805176765568 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:09:21.429579 133720644257344 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/items/array_metadatas/process_0 I0421 15:09:47.740787 133721028036160 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 27.087793s (commit=26.629595s, array_metadata_write=0.458198s) I0421 15:09:47.741971 133720981898816 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 444.346 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 28 seconds) (per-host) I0421 15:09:47.742120 133720981898816 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 27.439503s. I0421 15:09:48.054459 133720981898816 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784160076845627, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0421 15:09:48.256351 133720981898816 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:09:48.436690 133721002870336 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784160076845627, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0421 15:09:48.682542 133720981898816 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:09:48.683344 133720981898816 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.587646s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/items I0421 15:09:48.684086 133720981898816 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0/items I0421 15:09:48.926153 133720981898816 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0 I0421 15:09:49.616209 133720981898816 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0`. I0421 15:09:49.616957 133720981898816 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 30.408277s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/0 I0421 15:09:49.617040 133720981898816 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 29.314425s. I0421 15:09:49.617240 133720971413056 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:09:49.617356 133720971413056 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:09:49.617424 133720971413056 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:09:49.617471 133720971413056 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 15:09:49.617664 133805176765568 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. W0421 15:09:49.617977 133805176765568 checkpoint_manager.py:1441] Waiting for previous save to complete took 28.307962 seconds. If this number is high, consider checkpointing less frequently. I0421 15:09:49.618995 133805176765568 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0421 15:09:49.619330 133805176765568 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5. I0421 15:09:49.791724 133720971413056 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5 I0421 15:09:49.799140 133805176765568 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:09:49.799250 133805176765568 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 I0421 15:09:50.515713 133721028036160 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5/items I0421 15:10:05.310222 133805176765568 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 15.511575s I0421 15:10:05.314775 133805176765568 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 809.466 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 15 seconds) (per-host) I0421 15:10:05.314886 133805176765568 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 15.611414s (batch_requests_ready=0.091703s, total_serialization_initiated=15.515247s, others=0.004464s) I0421 15:10:05.314963 133805176765568 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 15.612016s (all_items=0.000023s, per_item={'items': '0.00002289'}, temp_paths=15.611993) I0421 15:10:05.316519 133720950441536 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:10:05.316603 133805176765568 async_checkpointer.py:561] Finished blocking save. Time taken: 15.697562s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5. I0421 15:10:05.316799 133805176765568 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:10:05.317029 133720843486784 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:10:05.317166 133805176765568 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784161.3099868, 'wait_for_prev_duration_secs': 28.30796194076538, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784189.6190193, 'checkpointer_blocking_duration_secs': 15.697689294815063, 'get_old_steps_start_time': 1776784205.3167324, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776784161.3099172, 'checkpoint_manager_blocking_duration_secs': 44.007219076156616} I0421 15:10:05.317325 133805176765568 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0421 15:10:05.318034 133805176765568 metric_logger.py:196] completed step: 4, seconds: 0.005, TFLOP/s/device: 2645.984, Tokens/s/device: 398831.548, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0421 15:10:05.326381 133805176765568 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.431, Tokens/s/device: 17248.349, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0421 15:10:05.442668 133805176765568 metric_logger.py:196] completed step: 6, seconds: 44.247, TFLOP/s/device: 0.307, Tokens/s/device: 46.286, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0421 15:10:05.561659 133805176765568 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 2120.008, Tokens/s/device: 319550.632, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0421 15:10:05.680459 133805176765568 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.812, Tokens/s/device: 17456.380, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0421 15:10:05.798057 133805176765568 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0421 15:10:05.798906 133805176765568 checkpointing.py:776] Waited 0.0008807182312011719 seconds for step 9 to finish before starting checkpointing. I0421 15:10:05.799177 133805176765568 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:10:06.069354 133721017550400 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5/items/array_metadatas/process_0 I0421 15:10:43.372435 133721028036160 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.056980s (commit=37.641634s, array_metadata_write=0.415347s) I0421 15:10:43.373695 133720950441536 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 235.453 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 53 seconds) (per-host) I0421 15:10:43.373756 133720950441536 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.057111s. I0421 15:10:43.801486 133720950441536 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:10:44.208905 133720950441536 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:10:44.209117 133720950441536 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.544389s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5/items I0421 15:10:44.209842 133720950441536 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5/items I0421 15:10:44.450550 133720950441536 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5 I0421 15:10:45.113243 133720950441536 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5`. I0421 15:10:45.113911 133720950441536 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 55.494876s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5 I0421 15:10:45.113986 133720950441536 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.797343s. I0421 15:10:45.114230 133720843486784 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:10:45.114352 133720843486784 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:10:45.114426 133720843486784 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:10:45.114470 133720843486784 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0421 15:10:45.114630 133805176765568 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0421 15:10:45.114758 133805176765568 checkpoint_manager.py:1441] Waiting for previous save to complete took 39.315588 seconds. If this number is high, consider checkpointing less frequently. I0421 15:10:45.115025 133805176765568 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0421 15:10:45.115360 133805176765568 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9. I0421 15:10:45.286847 133720843486784 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9 I0421 15:10:45.293245 133805176765568 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:10:45.293348 133805176765568 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 I0421 15:10:46.019993 133720971413056 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9/items I0421 15:10:56.222986 133805176765568 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.930186s I0421 15:10:56.227551 133805176765568 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.119 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0421 15:10:56.227655 133805176765568 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.027456s (batch_requests_ready=0.088994s, total_serialization_initiated=10.933997s, others=0.004466s) I0421 15:10:56.227732 133805176765568 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.028028s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=11.028013) I0421 15:10:56.228998 133720981898816 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:10:56.229093 133805176765568 async_checkpointer.py:561] Finished blocking save. Time taken: 11.114005s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9. I0421 15:10:56.229267 133805176765568 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:10:56.229427 133720918984256 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:10:56.229538 133805176765568 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784205.7991524, 'wait_for_prev_duration_secs': 39.31558799743652, 'time_between_consecutive_saves_sec': 16.18166494369507, 'checkpointer_blocking_start_time': 1776784245.115068, 'checkpointer_blocking_duration_secs': 11.114127159118652, 'get_old_steps_start_time': 1776784256.2292118, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1776784205.799118, 'checkpoint_manager_blocking_duration_secs': 50.430397748947144} I0421 15:10:56.229659 133805176765568 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0421 15:10:56.229693 133805176765568 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:10:56.975687 133721680250432 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9/items/array_metadatas/process_0 I0421 15:11:42.103926 133720971413056 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 45.875765s (commit=45.396413s, array_metadata_write=0.479352s) I0421 15:11:42.105170 133720981898816 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 222.068 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 56 seconds) (per-host) I0421 15:11:42.105287 133720981898816 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.876151s. I0421 15:11:42.540848 133720981898816 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:11:42.949192 133720981898816 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:11:42.949923 133720981898816 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.569022s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9/items I0421 15:11:42.950692 133720981898816 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9/items I0421 15:11:43.197239 133720981898816 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9 I0421 15:11:43.898639 133720981898816 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9`. I0421 15:11:43.899379 133720981898816 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 58.784297s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9 I0421 15:11:43.899447 133720981898816 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 47.670313s. I0421 15:11:43.899613 133720918984256 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:11:43.899727 133720918984256 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:11:43.899784 133720918984256 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:11:43.899825 133720918984256 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 15:11:43.899994 133805176765568 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. I0421 15:11:43.900169 133805176765568 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:11:43.902098 133805176765568 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.226, Tokens/s/device: 17217.318, total_weights: 16384, loss: 4.189, lm_loss: 4.189, perplexity: 65.979 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops