2026-04-17 13:49:58.451615: 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 13:49:58.567961 133117950094464 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:50:50.712517 133117950094464 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:50:50.712743 133117950094464 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:50:50.712791 133117950094464 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 13:50:50.712824 133117950094464 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:50:50.712852 133117950094464 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:50:50.712996 133117950094464 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:50:50.811057 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:50:51.206158 133117950094464 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:50:51.206573 133117950094464 checkpointing.py:677] Setting up checkpoint logger... I0417 13:50:51.206624 133117950094464 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:50:51.206716 133117950094464 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:50:51.207760 133117950094464 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 0x7911593f8aa0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:50:52.887698 133117950094464 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:50:52.887941 133117950094464 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 0x790add103e90>}, handler_registry=None I0417 13:50:52.888352 133117950094464 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x790add103e90>` 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 13:50:52.888396 133117950094464 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x790adcd32900>` 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 13:50:52.888427 133117950094464 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 0x790add103e90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x790add103e90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x790adcd32900>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x790adcd32900>}). I0417 13:50:52.888741 133117950094464 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:50:52.888798 133117950094464 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x790adcd2e200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:50:53.290766 133011825755712 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433716347577857, 'commit_timestamp_nsecs': 1776433769601031088, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0417 13:50:53.380134 133011836241472 checkpoint_manager.py:1847] Missing metrics for step 5 I0417 13:50:53.470068 133117950094464 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints I0417 13:50:53.470589 133117950094464 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_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x790adcd32600> I0417 13:50:53.470685 133117950094464 checkpointing.py:301] Checkpoint manager created! W0417 13:50:53.488103 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:50:54.051569 133117950094464 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:50:54.051671 133117950094464 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:50:54.136085 133117950094464 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:50:54.136162 133117950094464 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:50:54.150315 133117950094464 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:50:54.150364 133117950094464 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:50:54.170891 133117950094464 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 13:50:54.170946 133117950094464 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:50:54.185250 133117950094464 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 13:50:54.185298 133117950094464 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:50:54.199449 133117950094464 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 13:50:54.199497 133117950094464 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:50:54.221119 133117950094464 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:50:54.221176 133117950094464 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:50:54.239383 133117950094464 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:50:54.239439 133117950094464 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:50:54.432182 133117950094464 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:50:54.432271 133117950094464 checkpointing.py:581] restoring from this run's directory step 9 I0417 13:50:54.600716 133117950094464 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/9. I0417 13:50:55.155584 496638 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:50:55.155649 496638 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:51:01.240617 133117950094464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.958 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:51:01.240862 133117950094464 checkpointer.py:318] Finished restoring checkpoint in 6.73 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/9. I0417 13:51:01.240929 133117950094464 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776433854.5144658, 'checkpointer_duration_secs': 6.726438045501709, 'checkpoint_manager_start_time': 1776433854.432476, 'checkpoint_manager_duration_secs': 6.808429718017578} [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 13:51:01.607581 133117950094464 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:51:01.607694 133117950094464 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 13:51:01.607735 133117950094464 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 13:51:01.607780 133117950094464 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 13:51:01.607820 133117950094464 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 13:51:01.607914 133117950094464 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 13:51:01.607967 133117950094464 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 13:51:01.608009 133117950094464 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 13:51:01.608041 133117950094464 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 13:51:01.608227 133117950094464 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 13:51:01.608268 133117950094464 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:51:01.608304 133117950094464 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:51:02.025036 133117950094464 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:51:02.025130 133117950094464 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:51:02.039465 133117950094464 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:51:02.039517 133117950094464 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:51:02.269295 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:51:02.336103 133117950094464 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 13:51:02.336642 133117950094464 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:51:02.336804 133117950094464 metric_logger.py:301] number parameters: 1.104 billion W0417 13:51:03.160240 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:51:03.247859 133117950094464 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0417 13:51:03.364182 133117950094464 checkpointing.py:776] Waited 0.11630392074584961 seconds for step 10 to finish before starting checkpointing. I0417 13:51:03.364630 133117950094464 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:51:03.364794 133117950094464 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0417 13:51:03.365142 133117950094464 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10. I0417 13:51:03.476677 133117950094464 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:51:03.555944 133117950094464 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:51:03.556032 133117950094464 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 13:51:03.572661 133010229823040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10 W0417 13:51:03.772117 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.797840 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.805241 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.809801 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.814096 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.818160 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.822182 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:03.826749 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:51:04.290867 133010219337280 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10/items I0417 13:51:04.440316 133010187880000 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433864193683147, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776433864193683147, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 13:51:10.066826 133117950094464 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.511299s I0417 13:51:10.071478 133117950094464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.872 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:51:10.071547 133117950094464 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.593650s (batch_requests_ready=0.074982s, total_serialization_initiated=6.514132s, others=0.004536s) I0417 13:51:10.071632 133117950094464 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.594489s (all_items=0.000028s, per_item={'items': '0.00002789'}, temp_paths=6.594461) I0417 13:51:10.073096 133010240308800 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:51:10.073179 133117950094464 async_checkpointer.py:561] Finished blocking save. Time taken: 6.708342s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10. I0417 13:51:10.073390 133117950094464 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:51:10.073587 133010208851520 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:51:10.073690 133117950094464 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433863.3646157, 'wait_for_prev_duration_secs': 4.076957702636719e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433863.3648145, 'checkpointer_blocking_duration_secs': 6.7084760665893555, 'get_old_steps_start_time': 1776433870.073307, 'get_old_steps_duration_secs': 3.457069396972656e-05, 'checkpoint_manager_blocking_start_time': 1776433863.364509, 'checkpoint_manager_blocking_duration_secs': 6.709153890609741} I0417 13:51:10.073816 133117950094464 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0417 13:51:10.073858 133117950094464 max_utils.py:750] Memstats: After params initialized: I0417 13:51:10.073905 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:51:10.073931 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:51:10.073957 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:51:10.073977 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:51:10.073995 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:51:10.074012 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:51:10.074028 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:51:10.074059 133117950094464 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:51:10.079845 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:10.086255 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:51:10.091302 494227 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:51:10.280182 133117950094464 metric_logger.py:196] completed step: 10, seconds: 0.911, TFLOP/s/device: 14.922, Tokens/s/device: 2249.249, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0417 13:51:10.395973 133117950094464 metric_logger.py:196] completed step: 11, seconds: 7.030, TFLOP/s/device: 1.933, Tokens/s/device: 291.308, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0417 13:51:10.514572 133117950094464 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2392.943, Tokens/s/device: 360690.384, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0417 13:51:10.633092 133117950094464 metric_logger.py:196] completed step: 13, seconds: 0.117, TFLOP/s/device: 116.613, Tokens/s/device: 17577.136, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0417 13:51:10.637957 133117950094464 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0417 13:51:10.844372 133010198365760 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_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0417 13:51:10.868460 133117950094464 checkpointing.py:776] Waited 0.23048710823059082 seconds for step 15 to finish before starting checkpointing. I0417 13:51:10.868841 133117950094464 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:52:00.148267 133010219337280 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 50.076034s (commit=49.609913s, array_metadata_write=0.466121s) I0417 13:52:00.149440 133010240308800 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 222.983 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 56 seconds) (per-host) I0417 13:52:00.149494 133010240308800 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 50.076270s. I0417 13:52:00.527241 133010240308800 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 13:52:00.749841 133010187880000 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433864193683147, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 13:52:00.973386 133010240308800 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:52:00.973554 133010240308800 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.574962s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10/items I0417 13:52:00.974122 133010240308800 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10/items I0417 13:52:01.233508 133010240308800 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10 I0417 13:52:01.922368 133010240308800 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_15_ocdbt_true/checkpoints/10`. I0417 13:52:01.922991 133010240308800 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 58.558161s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/10 I0417 13:52:01.923086 133010240308800 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.849863s. I0417 13:52:01.923285 133010208851520 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:52:01.923394 133010208851520 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:52:01.923459 133010208851520 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:52:01.923506 133010208851520 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 13:52:01.924329 133117950094464 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 13:52:01.924460 133117950094464 checkpoint_manager.py:1441] Waiting for previous save to complete took 51.055614 seconds. If this number is high, consider checkpointing less frequently. I0417 13:52:01.925277 133117950094464 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0417 13:52:01.925555 133117950094464 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15. I0417 13:52:02.098637 133010208851520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15 I0417 13:52:02.105243 133117950094464 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:52:02.105336 133117950094464 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 13:52:02.797683 133010177394240 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15/items I0417 13:52:04.293775 133117950094464 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.188965s I0417 13:52:04.294312 133117950094464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.406 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0417 13:52:04.294368 133117950094464 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.282946s (batch_requests_ready=0.089757s, total_serialization_initiated=2.192759s, others=0.000430s) I0417 13:52:04.294435 133117950094464 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.283459s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=2.283443) I0417 13:52:04.299610 133011783812672 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:52:04.299711 133117950094464 async_checkpointer.py:561] Finished blocking save. Time taken: 2.374388s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15. I0417 13:52:04.299907 133117950094464 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:52:04.300061 133010240308800 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:52:04.300136 133117950094464 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433870.8688176, 'wait_for_prev_duration_secs': 51.05561375617981, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433921.925301, 'checkpointer_blocking_duration_secs': 2.3745195865631104, 'get_old_steps_start_time': 1776433924.2998395, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776433870.8687549, 'checkpoint_manager_blocking_duration_secs': 53.431355476379395} I0417 13:52:04.300250 133117950094464 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0417 13:52:04.300926 133117950094464 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.039, Tokens/s/device: 17189.139, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0417 13:52:04.308992 133117950094464 metric_logger.py:196] completed step: 15, seconds: 0.118, TFLOP/s/device: 114.943, Tokens/s/device: 17325.393, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0417 13:52:04.424178 133117950094464 metric_logger.py:196] completed step: 16, seconds: 53.669, TFLOP/s/device: 0.253, Tokens/s/device: 38.160, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0417 13:52:04.543064 133117950094464 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2326.563, Tokens/s/device: 350684.932, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0417 13:52:04.661006 133117950094464 metric_logger.py:196] completed step: 18, seconds: 0.116, TFLOP/s/device: 117.082, Tokens/s/device: 17647.870, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0417 13:52:04.778866 133117950094464 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0417 13:52:04.779686 133117950094464 checkpointing.py:776] Waited 0.0008368492126464844 seconds for step 19 to finish before starting checkpointing. I0417 13:52:04.779921 133117950094464 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:52:05.062796 133011794298432 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_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0417 13:52:51.815879 133010177394240 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 47.516936s (commit=47.066665s, array_metadata_write=0.450270s) I0417 13:52:51.816986 133011783812672 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 253.723 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 49 seconds) (per-host) I0417 13:52:51.817099 133011783812672 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 47.517342s. I0417 13:52:52.274983 133011783812672 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 13:52:52.715513 133011783812672 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:52:52.716168 133011783812672 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.592644s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15/items I0417 13:52:52.716812 133011783812672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15/items I0417 13:52:52.985936 133011783812672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15 I0417 13:52:53.714065 133011783812672 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_15_ocdbt_true/checkpoints/15`. I0417 13:52:53.714692 133011783812672 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 51.789375s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/15 I0417 13:52:53.714767 133011783812672 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 49.415012s. I0417 13:52:53.714875 133010240308800 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:52:53.714926 133010240308800 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:52:53.714986 133010240308800 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:52:53.715025 133010240308800 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 13:52:53.715747 133117950094464 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 13:52:53.715862 133117950094464 checkpoint_manager.py:1441] Waiting for previous save to complete took 48.935948 seconds. If this number is high, consider checkpointing less frequently. I0417 13:52:53.716096 133117950094464 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0417 13:52:53.716382 133117950094464 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19. I0417 13:52:53.893796 133117950094464 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:52:53.893884 133117950094464 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 13:52:53.895774 133010240308800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19 I0417 13:52:54.636075 133010219337280 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19/items I0417 13:53:00.951515 133117950094464 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.058107s I0417 13:53:00.952033 133117950094464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.727 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0417 13:53:00.952096 133117950094464 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.146626s (batch_requests_ready=0.084642s, total_serialization_initiated=7.061571s, others=0.000414s) I0417 13:53:00.952161 133117950094464 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.147116s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=7.147101) I0417 13:53:00.957545 133011741869632 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:53:00.957714 133117950094464 async_checkpointer.py:561] Finished blocking save. Time taken: 7.241574s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19. I0417 13:53:00.957905 133117950094464 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:53:00.958032 133011783812672 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:53:00.958122 133117950094464 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433924.779897, 'wait_for_prev_duration_secs': 48.93594813346863, 'time_between_consecutive_saves_sec': 2.856375217437744, 'checkpointer_blocking_start_time': 1776433973.7161214, 'checkpointer_blocking_duration_secs': 7.241714000701904, 'get_old_steps_start_time': 1776433980.957849, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1776433924.779868, 'checkpoint_manager_blocking_duration_secs': 56.17823100090027} I0417 13:53:00.958214 133117950094464 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0417 13:53:00.958243 133117950094464 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:53:01.729235 133011794298432 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_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0417 13:53:34.200600 133011752355392 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.243622s (commit=32.780803s, array_metadata_write=0.462819s) I0417 13:53:34.201715 133011741869632 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 312.821 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0417 13:53:34.201816 133011741869632 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.244059s. I0417 13:53:34.665579 133011741869632 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 13:53:35.093281 133011741869632 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:53:35.093957 133011741869632 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.577002s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19/items I0417 13:53:35.094513 133011741869632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19/items I0417 13:53:35.341773 133011741869632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19 I0417 13:53:36.028992 133011741869632 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_15_ocdbt_true/checkpoints/19`. I0417 13:53:36.029616 133011741869632 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.313482s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_true/checkpoints/19 I0417 13:53:36.029680 133011741869632 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.071924s. I0417 13:53:36.029791 133011783812672 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:53:36.029840 133011783812672 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:53:36.029887 133011783812672 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:53:36.029927 133011783812672 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 13:53:36.030083 133117950094464 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 13:53:36.030227 133117950094464 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:53:36.032096 133117950094464 metric_logger.py:196] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.612, Tokens/s/device: 17275.557, 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