2026-04-17 14:01:47.133147: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0417 14:01:47.250798 138256177978496 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 14:02:25.299288 138256177978496 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 14:02:25.299399 138256177978496 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 14:02:25.299436 138256177978496 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0417 14:02:25.299461 138256177978496 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 14:02:25.299483 138256177978496 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 14:02:25.299550 138256177978496 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 14:02:25.389675 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:25.773254 138256177978496 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 14:02:25.773559 138256177978496 checkpointing.py:677] Setting up checkpoint logger... I0417 14:02:25.773604 138256177978496 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 14:02:25.773674 138256177978496 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 14:02:25.774219 138256177978496 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 0x7dbdaf700b00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 14:02:27.436442 138256177978496 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 14:02:27.436653 138256177978496 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 0x7db8fd5fac90>}, handler_registry=None I0417 14:02:27.436984 138256177978496 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db8fd5fac90>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0417 14:02:27.437026 138256177978496 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db7339365a0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0417 14:02:27.437070 138256177978496 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 0x7db8fd5fac90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db8fd5fac90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db7339365a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db7339365a0>}). I0417 14:02:27.437372 138256177978496 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 14:02:27.437429 138256177978496 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7db733932200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 14:02:27.839005 138150049482304 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434439869036026, 'commit_timestamp_nsecs': 1776434479704255093, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0417 14:02:27.926740 138150049482304 checkpoint_manager.py:1847] Missing metrics for step 9 I0417 14:02:27.982292 138256177978496 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints I0417 14:02:27.982790 138256177978496 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_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7db7576ec0e0> I0417 14:02:27.982888 138256177978496 checkpointing.py:301] Checkpoint manager created! W0417 14:02:27.999687 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:28.505782 138256177978496 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 14:02:28.505880 138256177978496 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:02:28.588582 138256177978496 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 14:02:28.588671 138256177978496 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:02:28.602783 138256177978496 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 14:02:28.602829 138256177978496 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:02:28.623075 138256177978496 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 14:02:28.623130 138256177978496 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:02:28.637298 138256177978496 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 14:02:28.637349 138256177978496 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:02:28.651508 138256177978496 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 14:02:28.651556 138256177978496 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:02:28.672973 138256177978496 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 14:02:28.673030 138256177978496 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:02:28.691214 138256177978496 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 14:02:28.691265 138256177978496 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 14:02:28.883225 138256177978496 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 14:02:28.883316 138256177978496 checkpointing.py:581] restoring from this run's directory step 9 I0417 14:02:29.062776 138256177978496 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/9. I0417 14:02:29.602805 513880 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 14:02:29.602869 513880 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 14:02:35.653568 138256177978496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.968 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 14:02:35.653819 138256177978496 checkpointer.py:318] Finished restoring checkpoint in 6.67 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/9. I0417 14:02:35.653892 138256177978496 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776434548.9831088, 'checkpointer_duration_secs': 6.670755386352539, 'checkpoint_manager_start_time': 1776434548.8835244, 'checkpoint_manager_duration_secs': 6.770341873168945} [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0417 14:02:36.118106 138256177978496 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 14:02:36.118212 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 14:02:36.118253 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 14:02:36.118298 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 14:02:36.118338 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 14:02:36.118369 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 14:02:36.118408 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 14:02:36.118447 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0417 14:02:36.118479 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0417 14:02:36.118505 138256177978496 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 14:02:36.118539 138256177978496 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 14:02:36.118573 138256177978496 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 14:02:36.537620 138256177978496 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 14:02:36.537702 138256177978496 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 14:02:36.553289 138256177978496 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 14:02:36.553349 138256177978496 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 14:02:36.777091 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:36.841269 138256177978496 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0417 14:02:36.841845 138256177978496 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 14:02:36.842003 138256177978496 metric_logger.py:301] number parameters: 1.104 billion W0417 14:02:37.651013 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:37.720729 138256177978496 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0417 14:02:37.837180 138256177978496 checkpointing.py:776] Waited 0.11642837524414062 seconds for step 10 to finish before starting checkpointing. I0417 14:02:37.837584 138256177978496 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 14:02:37.837751 138256177978496 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0417 14:02:37.838039 138256177978496 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10. I0417 14:02:37.930168 138256177978496 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 14:02:38.014533 138256177978496 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:02:38.014623 138256177978496 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 14:02:38.020371 138148535338560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10 W0417 14:02:38.227531 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.252171 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.256560 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.260581 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.265062 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.268888 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.272715 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:38.276479 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:38.728297 138148524852800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/items I0417 14:02:38.916625 138148493395520 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434558640225542, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776434558640225542, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0417 14:02:46.028077 138256177978496 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.013928s I0417 14:02:46.032569 138256177978496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.523 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0417 14:02:46.032635 138256177978496 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.101095s (batch_requests_ready=0.079996s, total_serialization_initiated=8.016763s, others=0.004336s) I0417 14:02:46.032713 138256177978496 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.102107s (all_items=0.000038s, per_item={'items': '0.00003839'}, temp_paths=8.102069) I0417 14:02:46.034304 138148514367040 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:02:46.034415 138256177978496 async_checkpointer.py:561] Finished blocking save. Time taken: 8.196623s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10. I0417 14:02:46.034628 138256177978496 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:02:46.034796 138148545824320 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:02:46.034896 138256177978496 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434557.837572, 'wait_for_prev_duration_secs': 5.507469177246094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434557.837773, 'checkpointer_blocking_duration_secs': 8.196754693984985, 'get_old_steps_start_time': 1776434566.0345595, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776434557.8374898, 'checkpoint_manager_blocking_duration_secs': 8.197378158569336} I0417 14:02:46.035020 138256177978496 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0417 14:02:46.035079 138256177978496 max_utils.py:750] Memstats: After params initialized: I0417 14:02:46.035125 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 14:02:46.035151 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 14:02:46.035170 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 14:02:46.035189 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 14:02:46.035206 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 14:02:46.035223 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 14:02:46.035240 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 14:02:46.035257 138256177978496 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 14:02:46.041335 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:46.047425 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:02:46.051814 512282 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:02:46.239597 138256177978496 metric_logger.py:196] completed step: 10, seconds: 0.878, TFLOP/s/device: 15.471, Tokens/s/device: 2332.006, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0417 14:02:46.356155 138256177978496 metric_logger.py:196] completed step: 11, seconds: 8.517, TFLOP/s/device: 1.595, Tokens/s/device: 240.458, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0417 14:02:46.474534 138256177978496 metric_logger.py:196] completed step: 12, seconds: 0.005, TFLOP/s/device: 2536.805, Tokens/s/device: 382374.907, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0417 14:02:46.593462 138256177978496 metric_logger.py:196] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.683, Tokens/s/device: 17437.059, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0417 14:02:46.598200 138256177978496 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0417 14:02:46.769979 138148503881280 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0 I0417 14:02:46.830080 138256177978496 checkpointing.py:776] Waited 0.23182082176208496 seconds for step 15 to finish before starting checkpointing. I0417 14:02:46.830465 138256177978496 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:03:23.538060 138148524852800 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.504699s (commit=37.059406s, array_metadata_write=0.445294s) I0417 14:03:23.539190 138148514367040 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 277.076 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0417 14:03:23.539288 138148514367040 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.504830s. I0417 14:03:23.948426 138148514367040 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 14:03:24.154965 138148493395520 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434558640225542, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0417 14:03:24.389856 138148514367040 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 14:03:24.390455 138148514367040 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573499s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/items I0417 14:03:24.391132 138148514367040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10/items I0417 14:03:24.635027 138148514367040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10 I0417 14:03:25.303513 138148514367040 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10`. I0417 14:03:25.304156 138148514367040 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.466370s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/10 I0417 14:03:25.304229 138148514367040 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.269773s. I0417 14:03:25.304411 138148545824320 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:03:25.304518 138148545824320 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:03:25.304573 138148545824320 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:03:25.304618 138148545824320 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 14:03:25.304756 138256177978496 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. W0417 14:03:25.304887 138256177978496 checkpoint_manager.py:1441] Waiting for previous save to complete took 38.474416 seconds. If this number is high, consider checkpointing less frequently. I0417 14:03:25.305635 138256177978496 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0417 14:03:25.305921 138256177978496 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15. I0417 14:03:25.488013 138256177978496 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:03:25.488111 138256177978496 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 14:03:25.499433 138148545824320 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15 I0417 14:03:26.188252 138148482909760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15/items I0417 14:03:36.397260 138256177978496 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.909623s I0417 14:03:36.397870 138256177978496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.122 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0417 14:03:36.397936 138256177978496 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.000155s (batch_requests_ready=0.086648s, total_serialization_initiated=10.913012s, others=0.000496s) I0417 14:03:36.398021 138256177978496 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.000674s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=11.000657) I0417 14:03:36.403106 138148482909760 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:03:36.403211 138256177978496 async_checkpointer.py:561] Finished blocking save. Time taken: 11.097530s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15. I0417 14:03:36.403485 138256177978496 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:03:36.403630 138148514367040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:03:36.403708 138256177978496 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434566.8304381, 'wait_for_prev_duration_secs': 38.47441554069519, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434605.3056614, 'checkpointer_blocking_duration_secs': 11.097728252410889, 'get_old_steps_start_time': 1776434616.4034102, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776434566.8303776, 'checkpoint_manager_blocking_duration_secs': 49.57330322265625} I0417 14:03:36.403846 138256177978496 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0417 14:03:36.404601 138256177978496 metric_logger.py:196] completed step: 14, seconds: 0.118, TFLOP/s/device: 114.808, Tokens/s/device: 17305.044, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0417 14:03:36.412140 138256177978496 metric_logger.py:196] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.094, Tokens/s/device: 17197.511, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0417 14:03:36.528877 138256177978496 metric_logger.py:196] completed step: 16, seconds: 49.812, TFLOP/s/device: 0.273, Tokens/s/device: 41.115, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0417 14:03:36.647576 138256177978496 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2331.754, Tokens/s/device: 351467.307, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0417 14:03:36.766109 138256177978496 metric_logger.py:196] completed step: 18, seconds: 0.117, TFLOP/s/device: 115.786, Tokens/s/device: 17452.513, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0417 14:03:36.883473 138256177978496 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0417 14:03:36.884212 138256177978496 checkpointing.py:776] Waited 0.0007593631744384766 seconds for step 19 to finish before starting checkpointing. I0417 14:03:36.884449 138256177978496 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:03:37.153619 138148011050560 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0 I0417 14:04:14.260767 138150018025024 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.858347s (commit=37.407448s, array_metadata_write=0.450899s) I0417 14:04:14.261987 138148482909760 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 258.611 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0417 14:04:14.262126 138148482909760 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.858818s. I0417 14:04:14.666357 138148482909760 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 14:04:15.097968 138148482909760 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 14:04:15.098613 138148482909760 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.578911s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15/items I0417 14:04:15.099299 138148482909760 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15/items I0417 14:04:15.338321 138148482909760 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15 I0417 14:04:16.036076 138148482909760 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15`. I0417 14:04:16.036684 138148482909760 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.731011s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/15 I0417 14:04:16.036757 138148482909760 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.633452s. I0417 14:04:16.036935 138148514367040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:04:16.037055 138148514367040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:04:16.037136 138148514367040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:04:16.037181 138148514367040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 14:04:16.037984 138256177978496 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. W0417 14:04:16.038066 138256177978496 checkpoint_manager.py:1441] Waiting for previous save to complete took 39.153610 seconds. If this number is high, consider checkpointing less frequently. I0417 14:04:16.038278 138256177978496 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0417 14:04:16.038526 138256177978496 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19. I0417 14:04:16.223610 138256177978496 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:04:16.223700 138256177978496 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 14:04:16.228097 138148514367040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19 I0417 14:04:16.934471 138149965596224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19/items I0417 14:04:21.541199 138256177978496 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.317980s I0417 14:04:21.541721 138256177978496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.283 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0417 14:04:21.541775 138256177978496 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.406441s (batch_requests_ready=0.084575s, total_serialization_initiated=5.321458s, others=0.000407s) I0417 14:04:21.541843 138256177978496 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.406956s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=5.406942) I0417 14:04:21.547617 138148514367040 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:04:21.547702 138256177978496 async_checkpointer.py:561] Finished blocking save. Time taken: 5.509378s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19. I0417 14:04:21.547898 138256177978496 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:04:21.548060 138148482909760 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:04:21.548161 138256177978496 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434616.8844266, 'wait_for_prev_duration_secs': 39.15360999107361, 'time_between_consecutive_saves_sec': 11.579793691635132, 'checkpointer_blocking_start_time': 1776434656.0383017, 'checkpointer_blocking_duration_secs': 5.509508371353149, 'get_old_steps_start_time': 1776434661.5478308, 'get_old_steps_duration_secs': 2.4557113647460938e-05, 'checkpoint_manager_blocking_start_time': 1776434616.8843954, 'checkpoint_manager_blocking_duration_secs': 44.66373872756958} I0417 14:04:21.548296 138256177978496 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0417 14:04:21.548329 138256177978496 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:04:22.286615 138150007539264 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0 I0417 14:04:53.608392 138149976081984 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.061920s (commit=31.596582s, array_metadata_write=0.465338s) I0417 14:04:53.609433 138148514367040 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 337.214 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0417 14:04:53.609488 138148514367040 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.061743s. I0417 14:04:54.034969 138148514367040 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 14:04:54.518213 138148514367040 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 14:04:54.518381 138148514367040 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.625022s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19/items I0417 14:04:54.519077 138148514367040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19/items I0417 14:04:54.758654 138148514367040 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19 I0417 14:04:55.446623 138148514367040 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19`. I0417 14:04:55.447267 138148514367040 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.408952s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_true/checkpoints/19 I0417 14:04:55.447335 138148514367040 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.899589s. I0417 14:04:55.447526 138148482909760 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:04:55.447632 138148482909760 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:04:55.447684 138148482909760 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:04:55.447724 138148482909760 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 14:04:55.448307 138256177978496 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. I0417 14:04:55.448448 138256177978496 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 14:04:55.449569 138256177978496 metric_logger.py:196] completed step: 19, seconds: 0.118, TFLOP/s/device: 114.777, Tokens/s/device: 17300.366, 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