2026-04-21 16:04:22.938272: 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 00:00:1776787462.949582 1549180 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776787462.952941 1549180 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776787462.963930 1549180 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776787462.963942 1549180 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776787462.963944 1549180 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776787462.963946 1549180 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 16:04:26.336790: 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 16:04:26.452470 125316462865536 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( I0421 16:05:31.802644 125316462865536 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 16:05:31.802931 125316462865536 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 16:05:31.802979 125316462865536 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 16:05:31.803007 125316462865536 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 16:05:31.803029 125316462865536 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 16:05:31.803202 125316462865536 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 16:05:32.163170 125316462865536 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 16:05:32.163531 125316462865536 checkpointing.py:677] Setting up checkpoint logger... I0421 16:05:32.163580 125316462865536 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0421 16:05:32.163664 125316462865536 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 16:05:32.164466 125316462865536 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x71f9399adf10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 16:05:34.539593 125316462865536 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 16:05:34.539840 125316462865536 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 0x71f7e6fe6c00>}, handler_registry=None I0421 16:05:34.540240 125316462865536 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71f7e6fe6c00>` 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 16:05:34.540285 125316462865536 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f7e29505c0>` 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 16:05:34.540316 125316462865536 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 0x71f7e6fe6c00>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71f7e6fe6c00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f7e29505c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71f7e29505c0>}). I0421 16:05:34.540599 125316462865536 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 16:05:34.540668 125316462865536 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x71f7917ad300> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 16:05:34.669011 125316462865536 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints I0421 16:05:34.669285 125316462865536 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_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x71f791baeae0> I0421 16:05:34.669377 125316462865536 checkpointing.py:301] Checkpoint manager created! I0421 16:05:35.300225 125316462865536 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 16:05:35.300309 125316462865536 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 16:05:35.593309 125316462865536 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 16:05:35.593385 125316462865536 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 16:05:35.605330 125316462865536 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 16:05:35.605378 125316462865536 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 16:05:35.624200 125316462865536 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 16:05:35.624254 125316462865536 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 16:05:35.636278 125316462865536 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 16:05:35.636326 125316462865536 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 16:05:35.648209 125316462865536 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 16:05:35.648255 125316462865536 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 16:05:35.667261 125316462865536 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 16:05:35.667314 125316462865536 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 16:05:35.682459 125316462865536 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 16:05:35.682508 125316462865536 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 16:05:35.830309 125316462865536 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 16:05:35.830398 125316462865536 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 16:05:36.390863 125316462865536 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 16:05:36.390976 125316462865536 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 16:05:36.391018 125316462865536 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 16:05:36.391079 125316462865536 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 16:05:36.391123 125316462865536 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 16:05:36.391152 125316462865536 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 16:05:36.391191 125316462865536 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 16:05:36.391230 125316462865536 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 16:05:36.391260 125316462865536 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 16:05:36.391286 125316462865536 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 16:05:36.391320 125316462865536 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 16:05:36.391353 125316462865536 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 16:05:36.792179 125316462865536 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 16:05:36.792258 125316462865536 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 16:05:36.803618 125316462865536 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 16:05:36.803665 125316462865536 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 16:05:37.537647 125316462865536 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 16:05:37.538199 125316462865536 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 16:05:37.538348 125316462865536 metric_logger.py:301] number parameters: 1.104 billion I0421 16:05:39.308271 125316462865536 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0421 16:05:39.425775 125316462865536 checkpointing.py:776] Waited 0.11749124526977539 seconds for step 0 to finish before starting checkpointing. I0421 16:05:39.426228 125316462865536 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 16:05:39.426374 125316462865536 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0421 16:05:39.426745 125316462865536 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0. I0421 16:05:39.510798 125316462865536 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 16:05:39.591259 125232306193984 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0 I0421 16:05:39.606365 125316462865536 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 16:05:39.606454 125316462865536 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 16:05:40.338712 125232295708224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/items I0421 16:05:40.443657 125232251668032 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776787540199141590, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776787540199141590, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0421 16:05:40.701583 125316462865536 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.095620s I0421 16:05:40.701980 125316462865536 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 10.369 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0421 16:05:40.702029 125316462865536 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.190282s (batch_requests_ready=0.090852s, total_serialization_initiated=1.099121s, others=0.000309s) I0421 16:05:40.702102 125316462865536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.190825s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=1.190807) I0421 16:05:40.702810 125232230696512 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 16:05:40.702901 125316462865536 async_checkpointer.py:561] Finished blocking save. Time taken: 1.276490s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0. I0421 16:05:40.703101 125316462865536 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 16:05:40.703315 125232316679744 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 16:05:40.703417 125316462865536 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776787539.4262156, 'wait_for_prev_duration_secs': 3.933906555175781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776787539.4263937, 'checkpointer_blocking_duration_secs': 1.276623249053955, 'get_old_steps_start_time': 1776787540.7030323, 'get_old_steps_duration_secs': 3.695487976074219e-05, 'checkpoint_manager_blocking_start_time': 1776787539.426136, 'checkpoint_manager_blocking_duration_secs': 1.277256965637207} I0421 16:05:40.703531 125316462865536 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0421 16:05:40.703569 125316462865536 max_utils.py:750] Memstats: After params initialized: I0421 16:05:40.703631 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 16:05:40.703656 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 16:05:40.703675 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 16:05:40.703693 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 16:05:40.703710 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 16:05:40.703727 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 16:05:40.703743 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 16:05:40.703759 125316462865536 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 16:05:40.790086 1551259 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 16:05:40.790153 1551259 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 16:05:40.956033 125316462865536 metric_logger.py:196] completed step: 0, seconds: 1.769, TFLOP/s/device: 7.679, Tokens/s/device: 1157.454, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0421 16:05:40.956947 125316462865536 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/tensorboard/' I0421 16:05:41.072938 125316462865536 metric_logger.py:196] completed step: 1, seconds: 1.646, TFLOP/s/device: 8.255, Tokens/s/device: 1244.314, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0421 16:05:41.191755 125316462865536 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2332.154, Tokens/s/device: 351527.635, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0421 16:05:41.311028 125316462865536 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.655, Tokens/s/device: 17432.755, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0421 16:05:41.315942 125316462865536 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0421 16:05:41.548206 125316462865536 checkpointing.py:776] Waited 0.23223304748535156 seconds for step 5 to finish before starting checkpointing. I0421 16:05:41.548570 125316462865536 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 16:05:41.597038 125232306193984 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_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0421 16:06:22.331188 125232285222464 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.539136s (commit=41.095372s, array_metadata_write=0.443764s) I0421 16:06:22.332282 125232230696512 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 295.111 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 42 seconds) (per-host) I0421 16:06:22.332388 125232230696512 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.629433s. I0421 16:06:22.598305 125232230696512 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776787540199141590, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0421 16:06:22.783422 125232230696512 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 16:06:22.977339 125232251668032 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776787540199141590, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0421 16:06:23.307421 125232230696512 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.668592s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/items I0421 16:06:23.308160 125232230696512 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0/items I0421 16:06:23.562328 125232230696512 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0 I0421 16:06:24.255191 125232230696512 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_16_zarr3_false/checkpoints/0`. I0421 16:06:24.255805 125232230696512 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 44.829399s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/0 I0421 16:06:24.255886 125232230696512 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.552932s. I0421 16:06:24.256071 125232316679744 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 16:06:24.256181 125232316679744 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 16:06:24.256245 125232316679744 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 16:06:24.256291 125232316679744 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 16:06:24.256420 125316462865536 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 16:06:24.256540 125316462865536 checkpoint_manager.py:1441] Waiting for previous save to complete took 42.707968 seconds. If this number is high, consider checkpointing less frequently. I0421 16:06:24.257468 125316462865536 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0421 16:06:24.257759 125316462865536 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5. I0421 16:06:24.426885 125232316679744 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5 I0421 16:06:24.430616 125316462865536 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 16:06:24.430713 125316462865536 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 16:06:25.147197 125232163587648 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5/items I0421 16:06:33.529826 125316462865536 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.099718s I0421 16:06:33.534010 125316462865536 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.341 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0421 16:06:33.534267 125316462865536 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.201712s (batch_requests_ready=0.094497s, total_serialization_initiated=9.102978s, others=0.004237s) I0421 16:06:33.534385 125316462865536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.202247s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=9.202229) I0421 16:06:33.535673 125232316679744 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 16:06:33.535754 125316462865536 async_checkpointer.py:561] Finished blocking save. Time taken: 9.278241s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5. I0421 16:06:33.535965 125316462865536 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0421 16:06:33.536162 125232230696512 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 16:06:33.536288 125316462865536 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776787541.5485487, 'wait_for_prev_duration_secs': 42.70796847343445, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776787584.2574909, 'checkpointer_blocking_duration_secs': 9.278367042541504, 'get_old_steps_start_time': 1776787593.5358813, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776787541.5484874, 'checkpoint_manager_blocking_duration_secs': 51.987772941589355} I0421 16:06:33.536510 125316462865536 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0421 16:06:33.537224 125316462865536 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.853, Tokens/s/device: 17161.196, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0421 16:06:33.544880 125316462865536 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.164, Tokens/s/device: 17208.060, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0421 16:06:33.661390 125316462865536 metric_logger.py:196] completed step: 6, seconds: 52.227, TFLOP/s/device: 0.260, Tokens/s/device: 39.214, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0421 16:06:33.780318 125316462865536 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 2214.330, Tokens/s/device: 333767.927, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0421 16:06:33.898908 125316462865536 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.904, Tokens/s/device: 17470.378, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0421 16:06:34.017126 125316462865536 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0421 16:06:34.017938 125316462865536 checkpointing.py:776] Waited 0.0008263587951660156 seconds for step 9 to finish before starting checkpointing. I0421 16:06:34.018204 125316462865536 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 16:06:34.298206 125232306193984 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_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0421 16:07:12.134113 125232163587648 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.599372s (commit=38.166527s, array_metadata_write=0.432845s) I0421 16:07:12.135035 125232316679744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 264.354 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host) I0421 16:07:12.135097 125232316679744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.599302s. I0421 16:07:12.573288 125232316679744 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 16:07:13.052111 125232316679744 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.636409s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5/items I0421 16:07:13.052801 125232316679744 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5/items I0421 16:07:13.295394 125232316679744 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5 I0421 16:07:14.002622 125232316679744 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_16_zarr3_false/checkpoints/5`. I0421 16:07:14.003265 125232316679744 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.745759s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/5 I0421 16:07:14.003330 125232316679744 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.467536s. I0421 16:07:14.003482 125232230696512 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 16:07:14.003578 125232230696512 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 16:07:14.003638 125232230696512 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0421 16:07:14.003677 125232230696512 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0421 16:07:14.003815 125316462865536 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 16:07:14.003935 125316462865536 checkpoint_manager.py:1441] Waiting for previous save to complete took 39.985742 seconds. If this number is high, consider checkpointing less frequently. I0421 16:07:14.004160 125316462865536 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0421 16:07:14.004440 125316462865536 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9. I0421 16:07:14.196094 125232316679744 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9 I0421 16:07:14.207447 125316462865536 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 16:07:14.207543 125316462865536 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 16:07:14.890423 125232285222464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9/items I0421 16:07:29.214435 125316462865536 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 15.007313s I0421 16:07:29.218846 125316462865536 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 835.998 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 15 seconds) (per-host) I0421 16:07:29.218906 125316462865536 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 15.115896s (batch_requests_ready=0.101160s, total_serialization_initiated=15.010484s, others=0.004251s) I0421 16:07:29.218986 125316462865536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 15.116427s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=15.116414) I0421 16:07:29.220375 125233126180416 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 16:07:29.220456 125316462865536 async_checkpointer.py:561] Finished blocking save. Time taken: 15.216254s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9. I0421 16:07:29.220638 125316462865536 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 16:07:29.220845 125232306193984 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 16:07:29.220963 125316462865536 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776787594.018175, 'wait_for_prev_duration_secs': 39.98574161529541, 'time_between_consecutive_saves_sec': 9.76187014579773, 'checkpointer_blocking_start_time': 1776787634.0041847, 'checkpointer_blocking_duration_secs': 15.21637511253357, 'get_old_steps_start_time': 1776787649.2205763, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776787594.0181437, 'checkpoint_manager_blocking_duration_secs': 55.20278573036194} I0421 16:07:29.221120 125316462865536 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0421 16:07:29.221157 125316462865536 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 16:07:30.010752 125232132130368 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_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0421 16:08:15.492574 125232195044928 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 46.272830s (commit=45.838335s, array_metadata_write=0.434496s) I0421 16:08:15.493618 125233126180416 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 205.842 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host) I0421 16:08:15.493717 125233126180416 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.273218s. I0421 16:08:15.933425 125233126180416 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 16:08:16.429497 125233126180416 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.649558s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9/items I0421 16:08:16.430260 125233126180416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9/items I0421 16:08:16.693329 125233126180416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9 I0421 16:08:17.418373 125233126180416 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_16_zarr3_false/checkpoints/9`. I0421 16:08:17.418994 125233126180416 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 63.414798s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_16_zarr3_false/checkpoints/9 I0421 16:08:17.419089 125233126180416 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 48.198591s. I0421 16:08:17.419262 125232306193984 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 16:08:17.419369 125232306193984 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 16:08:17.419422 125232306193984 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 16:08:17.419463 125232306193984 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 16:08:17.419630 125316462865536 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 16:08:17.419765 125316462865536 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 16:08:17.420781 125316462865536 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 113.751, Tokens/s/device: 17145.823, 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