2026-04-21 15:12:04.377827: 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 00:00:1776784324.387823 1501256 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776784324.390897 1501256 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776784324.400845 1501256 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784324.400855 1501256 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784324.400857 1501256 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776784324.400859 1501256 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 15:12:07.848261: 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 15:12:07.965375 125494385298560 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:12:44.274814 125494385298560 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:12:44.275013 125494385298560 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:12:44.275071 125494385298560 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 15:12:44.275103 125494385298560 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:12:44.275125 125494385298560 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 15:12:44.275249 125494385298560 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:12:44.630612 125494385298560 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:12:44.630990 125494385298560 checkpointing.py:677] Setting up checkpoint logger... I0421 15:12:44.631039 125494385298560 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:12:44.631152 125494385298560 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:12:44.631981 125494385298560 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 0x7222a89af1a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:12:46.278963 125494385298560 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:12:46.279254 125494385298560 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 0x7220fbf726c0>}, handler_registry=None I0421 15:12:46.279641 125494385298560 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7220fbf726c0>` 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 15:12:46.279682 125494385298560 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7220fbf72cc0>` 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 15:12:46.279711 125494385298560 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 0x7220fbf726c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7220fbf726c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7220fbf72cc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7220fbf72cc0>}). I0421 15:12:46.280004 125494385298560 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:12:46.280086 125494385298560 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7220fbf991c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 15:12:46.684626 125410759149120 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784190425567961, 'commit_timestamp_nsecs': 1776784244496700500, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/5/_CHECKPOINT_METADATA I0421 15:12:46.771331 125410759149120 checkpoint_manager.py:1847] Missing metrics for step 5 I0421 15:12:46.827886 125494385298560 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints I0421 15:12:46.828493 125494385298560 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 0x7220fbf71610> I0421 15:12:46.828592 125494385298560 checkpointing.py:301] Checkpoint manager created! I0421 15:12:47.468938 125494385298560 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:12:47.469072 125494385298560 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:12:47.749655 125494385298560 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:12:47.749740 125494385298560 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:12:47.762085 125494385298560 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:12:47.762135 125494385298560 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:12:47.781177 125494385298560 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 15:12:47.781234 125494385298560 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:12:47.793579 125494385298560 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 15:12:47.793630 125494385298560 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:12:47.805827 125494385298560 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 15:12:47.805877 125494385298560 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:12:47.825219 125494385298560 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:12:47.825277 125494385298560 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:12:47.840624 125494385298560 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:12:47.840679 125494385298560 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:12:47.990462 125494385298560 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:12:47.990573 125494385298560 checkpointing.py:581] restoring from this run's directory step 9 I0421 15:12:48.145664 125494385298560 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9. I0421 15:12:48.716992 1502728 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:12:48.717079 1502728 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:12:54.509887 125494385298560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 2.046 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0421 15:12:54.510174 125494385298560 checkpointer.py:318] Finished restoring checkpoint in 6.44 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/9. I0421 15:12:54.510244 125494385298560 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776784368.06667, 'checkpointer_duration_secs': 6.443549156188965, 'checkpoint_manager_start_time': 1776784367.9907796, 'checkpoint_manager_duration_secs': 6.5194411277771} [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 15:12:54.885032 125494385298560 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:12:54.885159 125494385298560 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 15:12:54.885203 125494385298560 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 15:12:54.885248 125494385298560 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 15:12:54.885288 125494385298560 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 15:12:54.885316 125494385298560 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 15:12:54.885355 125494385298560 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 15:12:54.885394 125494385298560 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 15:12:54.885429 125494385298560 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 15:12:54.885457 125494385298560 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 15:12:54.885492 125494385298560 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:12:54.885528 125494385298560 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:12:55.538742 125494385298560 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:12:55.538865 125494385298560 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:12:55.550315 125494385298560 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:12:55.550365 125494385298560 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:12:56.191082 125494385298560 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 15:12:56.194615 125494385298560 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:12:56.194789 125494385298560 metric_logger.py:301] number parameters: 1.104 billion I0421 15:12:58.503014 125494385298560 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 15:12:58.619281 125494385298560 checkpointing.py:776] Waited 0.11624002456665039 seconds for step 10 to finish before starting checkpointing. I0421 15:12:58.619732 125494385298560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:12:58.619923 125494385298560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0421 15:12:58.620254 125494385298560 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/10. I0421 15:12:58.699642 125494385298560 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 15:12:58.773882 125410100643392 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10 I0421 15:12:58.800816 125494385298560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:12:58.800924 125494385298560 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 15:12:59.478187 125410090157632 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10/items I0421 15:12:59.625762 125410058700352 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784379390277429, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776784379390277429, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0421 15:13:06.041727 125494385298560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.241337s I0421 15:13:06.046481 125494385298560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.680 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0421 15:13:06.046551 125494385298560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.345907s (batch_requests_ready=0.096943s, total_serialization_initiated=7.244347s, others=0.004617s) I0421 15:13:06.046634 125494385298560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.346439s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=7.346418) I0421 15:13:06.048271 125410048214592 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:13:06.048357 125494385298560 async_checkpointer.py:561] Finished blocking save. Time taken: 7.428392s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10. I0421 15:13:06.048557 125494385298560 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:13:06.048800 125410111129152 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:13:06.048916 125494385298560 standard_logger.py:34] {'step': 10, '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': 1776784378.6197195, 'wait_for_prev_duration_secs': 3.886222839355469e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784378.6199465, 'checkpointer_blocking_duration_secs': 7.428518056869507, 'get_old_steps_start_time': 1776784386.0484827, 'get_old_steps_duration_secs': 3.266334533691406e-05, 'checkpoint_manager_blocking_start_time': 1776784378.619633, 'checkpoint_manager_blocking_duration_secs': 7.429257869720459} I0421 15:13:06.049058 125494385298560 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 15:13:06.049099 125494385298560 max_utils.py:750] Memstats: After params initialized: I0421 15:13:06.049146 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:13:06.049170 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:13:06.049189 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:13:06.049207 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:13:06.049223 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:13:06.049239 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:13:06.049255 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:13:06.049272 125494385298560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:13:06.299524 125494385298560 metric_logger.py:196] completed step: 10, seconds: 2.308, TFLOP/s/device: 5.888, Tokens/s/device: 887.489, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0421 15:13:06.416853 125494385298560 metric_logger.py:196] completed step: 11, seconds: 7.795, TFLOP/s/device: 1.743, Tokens/s/device: 262.749, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0421 15:13:06.536100 125494385298560 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2357.649, Tokens/s/device: 355370.467, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0421 15:13:06.654865 125494385298560 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 114.788, Tokens/s/device: 17302.120, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0421 15:13:06.659636 125494385298560 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 15:13:06.803715 125410069186112 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/10/items/array_metadatas/process_0 I0421 15:13:06.892411 125494385298560 checkpointing.py:776] Waited 0.23273944854736328 seconds for step 15 to finish before starting checkpointing. I0421 15:13:06.892816 125494385298560 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:13:42.158595 125410090157632 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.111086s (commit=35.666222s, array_metadata_write=0.444864s) I0421 15:13:42.159796 125410048214592 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 290.774 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 43 seconds) (per-host) I0421 15:13:42.159854 125410048214592 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.111452s. I0421 15:13:42.573919 125410048214592 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 15:13:42.777677 125410058700352 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776784379390277429, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0421 15:13:43.008242 125410048214592 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:13:43.008880 125410048214592 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.572881s. 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/10/items I0421 15:13:43.009631 125410048214592 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10/items I0421 15:13:43.260690 125410048214592 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10 I0421 15:13:43.927309 125410048214592 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/10`. I0421 15:13:43.928228 125410048214592 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.308269s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/10 I0421 15:13:43.928307 125410048214592 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.879907s. I0421 15:13:43.928523 125410111129152 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:13:43.928651 125410111129152 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:13:43.928715 125410111129152 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:13:43.928765 125410111129152 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 15:13:43.929774 125494385298560 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0421 15:13:43.930930 125494385298560 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.038117 seconds. If this number is high, consider checkpointing less frequently. I0421 15:13:43.931745 125494385298560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0421 15:13:43.932065 125494385298560 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/15. I0421 15:13:44.111887 125494385298560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:13:44.111987 125494385298560 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 15:13:44.116763 125410111129152 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15 I0421 15:13:44.827434 125410090157632 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15/items I0421 15:13:46.496116 125494385298560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.384700s I0421 15:13:46.500569 125494385298560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.978 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0421 15:13:46.500629 125494385298560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.479006s (batch_requests_ready=0.086349s, total_serialization_initiated=2.388322s, others=0.004335s) I0421 15:13:46.500707 125494385298560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.479570s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=2.479553) I0421 15:13:46.502363 125410717206080 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:13:46.502449 125494385298560 async_checkpointer.py:561] Finished blocking save. Time taken: 2.570660s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15. I0421 15:13:46.502636 125494385298560 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:13:46.502824 125410048214592 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:13:46.502940 125494385298560 standard_logger.py:34] {'step': 15, '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': 1776784386.8927894, 'wait_for_prev_duration_secs': 37.03811693191528, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784423.9317687, 'checkpointer_blocking_duration_secs': 2.5707833766937256, 'get_old_steps_start_time': 1776784426.502572, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1776784386.8927243, 'checkpoint_manager_blocking_duration_secs': 39.610190868377686} I0421 15:13:46.503078 125494385298560 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0421 15:13:46.503830 125494385298560 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.116, Tokens/s/device: 17200.833, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0421 15:13:46.512237 125494385298560 metric_logger.py:196] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.379, Tokens/s/device: 17240.508, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0421 15:13:46.629254 125494385298560 metric_logger.py:196] completed step: 16, seconds: 39.850, TFLOP/s/device: 0.341, Tokens/s/device: 51.392, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0421 15:13:46.747918 125494385298560 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2208.212, Tokens/s/device: 332845.766, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0421 15:13:46.866928 125494385298560 metric_logger.py:196] completed step: 18, seconds: 0.118, TFLOP/s/device: 115.269, Tokens/s/device: 17374.632, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0421 15:13:46.984982 125494385298560 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 15:13:46.985743 125494385298560 checkpointing.py:776] Waited 0.0007832050323486328 seconds for step 19 to finish before starting checkpointing. I0421 15:13:46.985998 125494385298560 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:13:47.224941 125410727691840 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/15/items/array_metadatas/process_0 I0421 15:14:22.134783 125410090157632 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.633406s (commit=35.176929s, array_metadata_write=0.456478s) I0421 15:14:22.135981 125410717206080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 331.549 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0421 15:14:22.136039 125410717206080 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.633546s. I0421 15:14:22.536146 125410717206080 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 15:14:22.976405 125410717206080 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:14:22.977121 125410717206080 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.588932s. 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/15/items I0421 15:14:22.977901 125410717206080 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15/items I0421 15:14:23.230935 125410717206080 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15 I0421 15:14:23.941560 125410717206080 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/15`. I0421 15:14:23.942307 125410717206080 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.010525s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/15 I0421 15:14:23.942383 125410717206080 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.439891s. I0421 15:14:23.942599 125410048214592 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:14:23.942714 125410048214592 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:14:23.942778 125410048214592 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:14:23.942825 125410048214592 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 15:14:23.943582 125494385298560 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0421 15:14:23.943709 125494385298560 checkpoint_manager.py:1441] Waiting for previous save to complete took 36.957718 seconds. If this number is high, consider checkpointing less frequently. I0421 15:14:23.943959 125494385298560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0421 15:14:23.944297 125494385298560 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/19. I0421 15:14:24.112391 125494385298560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:14:24.112489 125494385298560 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 15:14:24.114264 125410048214592 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19 I0421 15:14:24.842103 125410685748800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19/items I0421 15:14:31.928808 125494385298560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.816889s I0421 15:14:31.929470 125494385298560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.561 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0421 15:14:31.929527 125494385298560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.904198s (batch_requests_ready=0.083227s, total_serialization_initiated=7.820454s, others=0.000516s) I0421 15:14:31.929597 125494385298560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.904758s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=7.904741) I0421 15:14:31.934960 125410090157632 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:14:31.935061 125494385298560 async_checkpointer.py:561] Finished blocking save. Time taken: 7.991045s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19. I0421 15:14:31.935242 125494385298560 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:14:31.935428 125410717206080 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:14:31.935549 125494385298560 standard_logger.py:34] {'step': 19, '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': 1776784426.9859746, 'wait_for_prev_duration_secs': 36.95771789550781, 'time_between_consecutive_saves_sec': 3.0571954250335693, 'checkpointer_blocking_start_time': 1776784463.9439847, 'checkpointer_blocking_duration_secs': 7.991185188293457, 'get_old_steps_start_time': 1776784471.9351838, 'get_old_steps_duration_secs': 2.2411346435546875e-05, 'checkpoint_manager_blocking_start_time': 1776784426.9859426, 'checkpoint_manager_blocking_duration_secs': 44.94957947731018} I0421 15:14:31.935672 125494385298560 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 15:14:31.935706 125494385298560 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:14:32.671983 125410727691840 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/19/items/array_metadatas/process_0 I0421 15:15:05.440972 125410685748800 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.506832s (commit=33.044798s, array_metadata_write=0.462034s) I0421 15:15:05.442218 125410090157632 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 305.112 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0421 15:15:05.442277 125410090157632 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.507168s. I0421 15:15:05.897710 125410090157632 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 15:15:06.333798 125410090157632 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:15:06.334552 125410090157632 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.584809s. 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/19/items I0421 15:15:06.335308 125410090157632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19/items I0421 15:15:06.581089 125410090157632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19 I0421 15:15:07.294883 125410090157632 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/19`. I0421 15:15:07.295648 125410090157632 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.351651s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_08_checkpoint_async_true/checkpoints/19 I0421 15:15:07.295717 125410090157632 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.360608s. I0421 15:15:07.295931 125410717206080 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:15:07.296069 125410717206080 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:15:07.296129 125410717206080 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:15:07.296173 125410717206080 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 15:15:07.296507 125494385298560 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0421 15:15:07.297009 125494385298560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:15:07.298130 125494385298560 metric_logger.py:196] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.299, Tokens/s/device: 17228.326, total_weights: 16384, loss: 1.449, lm_loss: 1.449, perplexity: 4.260 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops