2026-04-21 15:46:46.791204: 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 00:00:1776786406.802317 1535722 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776786406.805494 1535722 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776786406.816376 1535722 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786406.816388 1535722 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786406.816390 1535722 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786406.816391 1535722 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 15:46:50.410076: 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 15:46:50.526792 131211733318784 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:47:45.618220 131211733318784 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:47:45.618331 131211733318784 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:47:45.618365 131211733318784 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 15:47:45.618390 131211733318784 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:47:45.618413 131211733318784 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 15:47:45.618487 131211733318784 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:47:45.973142 131211733318784 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:47:45.973509 131211733318784 checkpointing.py:677] Setting up checkpoint logger... I0421 15:47:45.973558 131211733318784 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:47:45.973662 131211733318784 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:47:45.974201 131211733318784 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 0x7755d31aef90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:47:47.631561 131211733318784 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:47:47.631939 131211733318784 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 0x77542bff19d0>}, handler_registry=None I0421 15:47:47.632416 131211733318784 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x77542bff19d0>` 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 15:47:47.632467 131211733318784 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77542bb7a810>` 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 15:47:47.632498 131211733318784 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 0x77542bff19d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x77542bff19d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77542bb7a810>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77542bb7a810>}). I0421 15:47:47.632835 131211733318784 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:47:47.632915 131211733318784 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x77542bb99260> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 15:47:48.050287 131128100914752 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776786351523338407, 'commit_timestamp_nsecs': 1776786385547861725, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0421 15:47:48.137741 131128100914752 checkpoint_manager.py:1847] Missing metrics for step 9 I0421 15:47:48.226700 131211733318784 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints I0421 15:47:48.227254 131211733318784 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_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x77542bb79370> I0421 15:47:48.227354 131211733318784 checkpointing.py:301] Checkpoint manager created! I0421 15:47:48.910776 131211733318784 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:47:48.910864 131211733318784 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:47:49.188885 131211733318784 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:47:49.188961 131211733318784 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:47:49.200905 131211733318784 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:47:49.200949 131211733318784 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:47:49.219529 131211733318784 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 15:47:49.219582 131211733318784 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:47:49.231564 131211733318784 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 15:47:49.231622 131211733318784 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:47:49.243552 131211733318784 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 15:47:49.243605 131211733318784 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:47:49.262517 131211733318784 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:47:49.262573 131211733318784 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:47:49.277748 131211733318784 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:47:49.277798 131211733318784 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:47:49.425668 131211733318784 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:47:49.425758 131211733318784 checkpointing.py:581] restoring from this run's directory step 9 I0421 15:47:49.601155 131211733318784 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/9. I0421 15:47:50.154314 1537240 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:47:50.154389 1537240 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:47:56.490124 131211733318784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.886 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0421 15:47:56.490391 131211733318784 checkpointer.py:318] Finished restoring checkpoint in 6.98 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/9. I0421 15:47:56.490470 131211733318784 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776786469.5140133, 'checkpointer_duration_secs': 6.976423740386963, 'checkpoint_manager_start_time': 1776786469.4259574, 'checkpoint_manager_duration_secs': 7.0644824504852295} [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 15:47:56.811164 131211733318784 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:47:56.811271 131211733318784 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 15:47:56.811313 131211733318784 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 15:47:56.811358 131211733318784 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 15:47:56.811398 131211733318784 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 15:47:56.811425 131211733318784 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 15:47:56.811465 131211733318784 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 15:47:56.811503 131211733318784 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 15:47:56.811534 131211733318784 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 15:47:56.811562 131211733318784 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 15:47:56.811597 131211733318784 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:47:56.811631 131211733318784 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:47:57.214192 131211733318784 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:47:57.214270 131211733318784 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:47:57.225900 131211733318784 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:47:57.225950 131211733318784 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:47:57.844563 131211733318784 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 15:47:57.848090 131211733318784 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:47:57.848257 131211733318784 metric_logger.py:301] number parameters: 1.104 billion I0421 15:47:59.853013 131211733318784 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 15:47:59.969554 131211733318784 checkpointing.py:776] Waited 0.1165304183959961 seconds for step 10 to finish before starting checkpointing. I0421 15:47:59.969954 131211733318784 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:47:59.970151 131211733318784 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0421 15:47:59.970446 131211733318784 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10. I0421 15:48:00.069051 131211733318784 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 15:48:00.147368 131211733318784 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:48:00.147457 131211733318784 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 15:48:00.155486 131127446603328 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10 I0421 15:48:00.863382 131127436117568 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10/items I0421 15:48:01.000625 131127404660288 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776786480768436236, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776786480768436236, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0421 15:48:08.389600 131211733318784 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.242619s I0421 15:48:08.394168 131211733318784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.483 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0421 15:48:08.394237 131211733318784 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.324284s (batch_requests_ready=0.074376s, total_serialization_initiated=8.245443s, others=0.004464s) I0421 15:48:08.394322 131211733318784 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.324832s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=8.324813) I0421 15:48:08.395586 131127446603328 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:48:08.395669 131211733318784 async_checkpointer.py:561] Finished blocking save. Time taken: 8.425477s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10. I0421 15:48:08.395869 131211733318784 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:48:08.396082 131127457089088 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:48:08.396185 131211733318784 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776786479.9699407, 'wait_for_prev_duration_secs': 3.910064697265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776786479.9701738, 'checkpointer_blocking_duration_secs': 8.42560338973999, 'get_old_steps_start_time': 1776786488.395796, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776786479.969859, 'checkpoint_manager_blocking_duration_secs': 8.42629861831665} I0421 15:48:08.396306 131211733318784 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 15:48:08.396347 131211733318784 max_utils.py:750] Memstats: After params initialized: I0421 15:48:08.396400 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:48:08.396426 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:48:08.396446 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:48:08.396466 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:48:08.396481 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:48:08.396497 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:48:08.396513 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:48:08.396529 131211733318784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:48:08.651269 131211733318784 metric_logger.py:196] completed step: 10, seconds: 2.004, TFLOP/s/device: 6.779, Tokens/s/device: 1021.848, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0421 15:48:08.768037 131211733318784 metric_logger.py:196] completed step: 11, seconds: 8.796, TFLOP/s/device: 1.545, Tokens/s/device: 232.824, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0421 15:48:08.886954 131211733318784 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2374.127, Tokens/s/device: 357854.272, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0421 15:48:09.006031 131211733318784 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 115.626, Tokens/s/device: 17428.452, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0421 15:48:09.010523 131211733318784 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 15:48:09.151565 131127415146048 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_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0421 15:48:09.243507 131211733318784 checkpointing.py:776] Waited 0.2329545021057129 seconds for step 15 to finish before starting checkpointing. I0421 15:48:09.243890 131211733318784 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:48:48.921855 131127436117568 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.527028s (commit=40.069560s, array_metadata_write=0.457469s) I0421 15:48:48.922962 131127446603328 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 258.670 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0421 15:48:48.923022 131127446603328 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.527310s. I0421 15:48:49.349186 131127446603328 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 15:48:49.551293 131127404660288 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776786480768436236, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0421 15:48:49.769100 131127446603328 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:48:49.769719 131127446603328 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570614s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10/items I0421 15:48:49.770358 131127446603328 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10/items I0421 15:48:50.032268 131127446603328 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10 I0421 15:48:50.709845 131127446603328 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_15_ocdbt_true/checkpoints/10`. I0421 15:48:50.710483 131127446603328 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.740298s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/10 I0421 15:48:50.710561 131127446603328 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 42.314850s. I0421 15:48:50.710745 131127457089088 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:48:50.710855 131127457089088 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:48:50.710927 131127457089088 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:48:50.710977 131127457089088 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 15:48:50.711570 131211733318784 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 15:48:50.711704 131211733318784 checkpoint_manager.py:1441] Waiting for previous save to complete took 41.467811 seconds. If this number is high, consider checkpointing less frequently. I0421 15:48:50.712470 131211733318784 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0421 15:48:50.712764 131211733318784 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15. I0421 15:48:50.886597 131127457089088 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15 I0421 15:48:50.901635 131211733318784 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:48:50.901725 131211733318784 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 15:48:51.615184 131127425631808 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15/items I0421 15:48:57.424866 131211733318784 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.523638s I0421 15:48:57.425399 131211733318784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.863 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0421 15:48:57.425455 131211733318784 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.623547s (batch_requests_ready=0.096328s, total_serialization_initiated=6.526799s, others=0.000420s) I0421 15:48:57.425524 131211733318784 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.624057s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=6.624042) I0421 15:48:57.431118 131128017028672 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:48:57.431206 131211733318784 async_checkpointer.py:561] Finished blocking save. Time taken: 6.718685s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15. I0421 15:48:57.431391 131211733318784 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:48:57.431595 131128006542912 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:48:57.431713 131211733318784 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776786489.2438648, 'wait_for_prev_duration_secs': 41.46781086921692, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776786530.7124927, 'checkpointer_blocking_duration_secs': 6.718801736831665, 'get_old_steps_start_time': 1776786537.4313204, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776786489.2438028, 'checkpoint_manager_blocking_duration_secs': 48.187883138656616} I0421 15:48:57.431871 131211733318784 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0421 15:48:57.432633 131211733318784 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 113.882, Tokens/s/device: 17165.511, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0421 15:48:57.441553 131211733318784 metric_logger.py:196] completed step: 15, seconds: 0.118, TFLOP/s/device: 114.670, Tokens/s/device: 17284.305, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0421 15:48:57.557373 131211733318784 metric_logger.py:196] completed step: 16, seconds: 48.429, TFLOP/s/device: 0.281, Tokens/s/device: 42.289, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0421 15:48:57.676630 131211733318784 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2129.310, Tokens/s/device: 320952.829, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0421 15:48:57.795367 131211733318784 metric_logger.py:196] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.308, Tokens/s/device: 17531.245, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0421 15:48:57.913275 131211733318784 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 15:48:57.914065 131211733318784 checkpointing.py:776] Waited 0.0007984638214111328 seconds for step 19 to finish before starting checkpointing. I0421 15:48:57.914309 131211733318784 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:48:58.158794 131128069457472 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_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0421 15:49:36.680451 131128058971712 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.250069s (commit=38.799138s, array_metadata_write=0.450931s) I0421 15:49:36.681525 131128017028672 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 275.434 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0421 15:49:36.681622 131128017028672 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.250453s. I0421 15:49:37.118305 131128017028672 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 15:49:37.567378 131128017028672 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:49:37.568031 131128017028672 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.595995s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15/items I0421 15:49:37.568671 131128017028672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15/items I0421 15:49:37.810232 131128017028672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15 I0421 15:49:38.482280 131128017028672 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_15_ocdbt_true/checkpoints/15`. I0421 15:49:38.482908 131128017028672 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.770401s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/15 I0421 15:49:38.482978 131128017028672 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.051811s. I0421 15:49:38.483162 131128006542912 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:49:38.483267 131128006542912 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:49:38.483334 131128006542912 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:49:38.483378 131128006542912 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 15:49:38.484331 131211733318784 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 15:49:38.484441 131211733318784 checkpoint_manager.py:1441] Waiting for previous save to complete took 40.570143 seconds. If this number is high, consider checkpointing less frequently. I0421 15:49:38.484654 131211733318784 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0421 15:49:38.484944 131211733318784 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19. I0421 15:49:38.658700 131128006542912 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19 I0421 15:49:38.664723 131211733318784 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:49:38.664813 131211733318784 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 15:49:39.401372 131127457089088 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19/items I0421 15:49:52.566780 131211733318784 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 13.902426s I0421 15:49:52.570806 131211733318784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 903.009 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 13 seconds) (per-host) I0421 15:49:52.570864 131211733318784 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 13.994175s (batch_requests_ready=0.084522s, total_serialization_initiated=13.905732s, others=0.003921s) I0421 15:49:52.570940 131211733318784 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 13.995025s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=13.995006) I0421 15:49:52.572475 131128058971712 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:49:52.572574 131211733318784 async_checkpointer.py:561] Finished blocking save. Time taken: 14.087879s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19. I0421 15:49:52.572816 131211733318784 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:49:52.572947 131128017028672 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:49:52.573022 131211733318784 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776786537.9142828, 'wait_for_prev_duration_secs': 40.5701425075531, 'time_between_consecutive_saves_sec': 7.20328950881958, 'checkpointer_blocking_start_time': 1776786578.484678, 'checkpointer_blocking_duration_secs': 14.088072299957275, 'get_old_steps_start_time': 1776786592.5727634, 'get_old_steps_duration_secs': 2.0503997802734375e-05, 'checkpoint_manager_blocking_start_time': 1776786537.9142516, 'checkpoint_manager_blocking_duration_secs': 54.658748388290405} I0421 15:49:52.573120 131211733318784 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 15:49:52.573149 131211733318784 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:49:53.303761 131132358133312 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_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0421 15:50:43.487152 131128069457472 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 50.915661s (commit=50.454871s, array_metadata_write=0.460790s) I0421 15:50:43.488274 131128058971712 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 194.677 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host) I0421 15:50:43.488380 131128058971712 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 50.915760s. I0421 15:50:43.940639 131128058971712 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 15:50:44.364963 131128058971712 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:50:44.365645 131128058971712 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.569756s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19/items I0421 15:50:44.366301 131128058971712 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19/items I0421 15:50:44.618558 131128058971712 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19 I0421 15:50:45.323426 131128058971712 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_15_ocdbt_true/checkpoints/19`. I0421 15:50:45.324030 131128058971712 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 66.839340s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_15_ocdbt_true/checkpoints/19 I0421 15:50:45.324116 131128058971712 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 52.751493s. I0421 15:50:45.324269 131128017028672 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:50:45.324318 131128017028672 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:50:45.324368 131128017028672 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:50:45.324404 131128017028672 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 15:50:45.324542 131211733318784 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 15:50:45.324682 131211733318784 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:50:45.325523 131211733318784 metric_logger.py:196] completed step: 19, seconds: 0.119, TFLOP/s/device: 113.908, Tokens/s/device: 17169.397, 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