2026-04-17 13:56:09.782309: 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 13:56:09.899919 123422613400704 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:57:03.959060 123422613400704 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:57:03.959166 123422613400704 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:57:03.959199 123422613400704 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 13:57:03.959228 123422613400704 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:57:03.959251 123422613400704 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:57:03.959366 123422613400704 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:57:04.052871 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:04.457656 123422613400704 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:57:04.458003 123422613400704 checkpointing.py:677] Setting up checkpoint logger... I0417 13:57:04.458061 123422613400704 checkpointing.py:233] Creating checkpoint manager with ocdbt=False and zarr3=True I0417 13:57:04.458116 123422613400704 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:57:04.458756 123422613400704 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x703ffabd39e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:57:06.086507 123422613400704 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:57:06.086692 123422613400704 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 0x70397e3320c0>}, handler_registry=None I0417 13:57:06.086983 123422613400704 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70397e3320c0>` 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 13:57:06.087025 123422613400704 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70397e332660>` 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 13:57:06.087067 123422613400704 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 0x70397e3320c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70397e3320c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70397e332660>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70397e332660>}). I0417 13:57:06.087347 123422613400704 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:57:06.087400 123422613400704 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x70397e32e200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:57:06.593536 123316513801792 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434104053089549, 'commit_timestamp_nsecs': 1776434110989095409, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0417 13:57:06.675939 123316513801792 checkpoint_manager.py:1847] Missing metrics for step 0 I0417 13:57:06.715369 123422613400704 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints I0417 13:57:06.715847 123422613400704 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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x70397e331490> I0417 13:57:06.715945 123422613400704 checkpointing.py:301] Checkpoint manager created! W0417 13:57:06.733013 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:07.271555 123422613400704 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:57:07.271656 123422613400704 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:57:07.354811 123422613400704 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:57:07.354893 123422613400704 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:57:07.368946 123422613400704 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:57:07.369002 123422613400704 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:57:07.389286 123422613400704 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 13:57:07.389344 123422613400704 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:57:07.403450 123422613400704 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 13:57:07.403497 123422613400704 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:57:07.417534 123422613400704 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 13:57:07.417582 123422613400704 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:57:07.439018 123422613400704 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:57:07.439088 123422613400704 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:57:07.457126 123422613400704 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:57:07.457178 123422613400704 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:57:07.648949 123422613400704 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:57:07.649057 123422613400704 checkpointing.py:581] restoring from this run's directory step 9 I0417 13:57:07.812317 123422613400704 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/9. I0417 13:57:08.631864 506058 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:57:08.631918 506058 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:57:14.616946 123422613400704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.920 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:57:14.617216 123422613400704 checkpointer.py:318] Finished restoring checkpoint in 6.88 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/9. I0417 13:57:14.617282 123422613400704 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776434227.7322059, 'checkpointer_duration_secs': 6.885052919387817, 'checkpoint_manager_start_time': 1776434227.6492712, 'checkpoint_manager_duration_secs': 6.967989206314087} [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 13:57:14.987792 123422613400704 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:57:14.987901 123422613400704 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 13:57:14.987942 123422613400704 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 13:57:14.987987 123422613400704 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 13:57:14.988029 123422613400704 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 13:57:14.988072 123422613400704 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 13:57:14.988114 123422613400704 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 13:57:14.988152 123422613400704 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 13:57:14.988183 123422613400704 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 13:57:14.988208 123422613400704 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 13:57:14.988242 123422613400704 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:57:14.988276 123422613400704 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:57:15.405141 123422613400704 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:57:15.405224 123422613400704 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:57:15.418732 123422613400704 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:57:15.418780 123422613400704 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:57:15.644568 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:15.722689 123422613400704 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 13:57:15.723256 123422613400704 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:57:15.723415 123422613400704 metric_logger.py:301] number parameters: 1.104 billion W0417 13:57:16.540913 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:16.610322 123422613400704 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0417 13:57:16.726535 123422613400704 checkpointing.py:776] Waited 0.11620140075683594 seconds for step 10 to finish before starting checkpointing. I0417 13:57:16.726928 123422613400704 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:57:16.727110 123422613400704 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0417 13:57:16.727391 123422613400704 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_false/checkpoints/10. I0417 13:57:16.811766 123422613400704 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:57:16.882810 123314974492224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10 I0417 13:57:16.901002 123422613400704 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:57:16.901106 123422613400704 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 W0417 13:57:17.138668 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.157277 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.161851 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.166124 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.171452 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.175410 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.179307 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:17.183405 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:17.534111 123314964006464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/items I0417 13:57:17.697732 123314932549184 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434237456506432, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776434237456506432, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0417 13:57:23.117345 123422613400704 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.216784s I0417 13:57:23.121599 123422613400704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.956 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:57:23.121916 123422613400704 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.309209s (batch_requests_ready=0.085201s, total_serialization_initiated=6.219660s, others=0.004349s) I0417 13:57:23.122010 123422613400704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.309782s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=6.309762) I0417 13:57:23.123223 123314953520704 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:57:23.123305 123422613400704 async_checkpointer.py:561] Finished blocking save. Time taken: 6.396152s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10. I0417 13:57:23.123514 123422613400704 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:57:23.123710 123314984977984 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:57:23.123806 123422613400704 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434236.726915, 'wait_for_prev_duration_secs': 3.910064697265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434236.7271335, 'checkpointer_blocking_duration_secs': 6.396282196044922, 'get_old_steps_start_time': 1776434243.1234457, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776434236.7268322, 'checkpoint_manager_blocking_duration_secs': 6.396945953369141} I0417 13:57:23.123942 123422613400704 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0417 13:57:23.123990 123422613400704 max_utils.py:750] Memstats: After params initialized: I0417 13:57:23.124035 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:57:23.124074 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:57:23.124094 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:57:23.124113 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:57:23.124130 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:57:23.124147 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:57:23.124164 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:57:23.124180 123422613400704 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:57:23.130756 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:23.137184 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:57:23.142166 503782 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:57:23.330570 123422613400704 metric_logger.py:196] completed step: 10, seconds: 0.886, TFLOP/s/device: 15.328, Tokens/s/device: 2310.443, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0417 13:57:23.446670 123422613400704 metric_logger.py:196] completed step: 11, seconds: 6.718, TFLOP/s/device: 2.022, Tokens/s/device: 304.841, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0417 13:57:23.565932 123422613400704 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2269.439, Tokens/s/device: 342074.495, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0417 13:57:23.684367 123422613400704 metric_logger.py:196] completed step: 13, seconds: 0.119, TFLOP/s/device: 114.624, Tokens/s/device: 17277.306, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0417 13:57:23.693763 123422613400704 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0417 13:57:23.898470 123314943034944 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_false/checkpoints/10/items/array_metadatas/process_0 I0417 13:57:23.921896 123422613400704 checkpointing.py:776] Waited 0.22809481620788574 seconds for step 15 to finish before starting checkpointing. I0417 13:57:23.922379 123422613400704 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:57:29.064941 123314964006464 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 5.942593s (commit=5.579334s, array_metadata_write=0.363259s) I0417 13:57:29.065899 123314953520704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1.007 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0417 13:57:29.066008 123314953520704 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 5.942657s. W0417 13:57:29.345937 123314953520704 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/items, I0417 13:57:29.476063 123314953520704 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 13:57:29.476173 123314953520704 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.175084s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/items I0417 13:57:29.476808 123314953520704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/items I0417 13:57:29.637335 123314932549184 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434237456506432, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0417 13:57:29.713609 123314953520704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10 I0417 13:57:30.392749 123314953520704 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_false/checkpoints/10`. I0417 13:57:30.393378 123314953520704 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 13.666231s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/10 I0417 13:57:30.393452 123314953520704 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 7.270102s. I0417 13:57:30.393630 123314984977984 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:57:30.393737 123314984977984 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:57:30.393797 123314984977984 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:57:30.393841 123314984977984 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 13:57:30.394301 123422613400704 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 13:57:30.394545 123422613400704 checkpoint_manager.py:1441] Waiting for previous save to complete took 6.472179 seconds. If this number is high, consider checkpointing less frequently. I0417 13:57:30.395213 123422613400704 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0417 13:57:30.395485 123422613400704 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_false/checkpoints/15. I0417 13:57:30.542453 123314984977984 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15 I0417 13:57:30.556850 123422613400704 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:57:30.556948 123422613400704 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 13:57:31.202163 123314974492224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15/items I0417 13:57:35.762449 123422613400704 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.205989s I0417 13:57:35.766438 123422613400704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.332 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0417 13:57:35.766507 123422613400704 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.292756s (batch_requests_ready=0.079906s, total_serialization_initiated=5.209024s, others=0.003826s) I0417 13:57:35.766589 123422613400704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.293489s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=5.293473) I0417 13:57:35.767996 123314984977984 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:57:35.768134 123422613400704 async_checkpointer.py:561] Finished blocking save. Time taken: 5.372859s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15. I0417 13:57:35.768341 123422613400704 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:57:35.768504 123314953520704 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:57:35.768616 123422613400704 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434243.9223475, 'wait_for_prev_duration_secs': 6.472179412841797, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434250.3952374, 'checkpointer_blocking_duration_secs': 5.373008966445923, 'get_old_steps_start_time': 1776434255.7682688, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776434243.9222748, 'checkpoint_manager_blocking_duration_secs': 11.84630298614502} I0417 13:57:35.768737 123422613400704 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0417 13:57:35.769365 123422613400704 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 113.809, Tokens/s/device: 17154.584, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0417 13:57:35.776669 123422613400704 metric_logger.py:196] completed step: 15, seconds: 0.121, TFLOP/s/device: 112.215, Tokens/s/device: 16914.297, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0417 13:57:35.891850 123422613400704 metric_logger.py:196] completed step: 16, seconds: 12.081, TFLOP/s/device: 1.125, Tokens/s/device: 169.524, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0417 13:57:36.010681 123422613400704 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2210.008, Tokens/s/device: 333116.461, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0417 13:57:36.129271 123422613400704 metric_logger.py:196] completed step: 18, seconds: 0.116, TFLOP/s/device: 117.536, Tokens/s/device: 17716.263, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0417 13:57:36.246984 123422613400704 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0417 13:57:36.248446 123422613400704 checkpointing.py:776] Waited 0.001505136489868164 seconds for step 19 to finish before starting checkpointing. I0417 13:57:36.248832 123422613400704 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:57:36.481626 123314943034944 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_false/checkpoints/15/items/array_metadatas/process_0 I0417 13:57:42.195704 123314974492224 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.428429s (commit=6.039650s, array_metadata_write=0.388778s) I0417 13:57:42.196683 123314984977984 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1.053 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0417 13:57:42.196802 123314984977984 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.428617s. W0417 13:57:42.498686 123314984977984 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15/items, I0417 13:57:42.590017 123314984977984 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 13:57:42.590143 123314984977984 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.161329s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15/items I0417 13:57:42.590775 123314984977984 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15/items I0417 13:57:42.842765 123314984977984 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15 I0417 13:57:43.475844 123314984977984 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_false/checkpoints/15`. I0417 13:57:43.476482 123314984977984 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 13.081231s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/15 I0417 13:57:43.476549 123314984977984 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 7.708367s. I0417 13:57:43.476708 123314953520704 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:57:43.476807 123314953520704 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:57:43.476871 123314953520704 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:57:43.476911 123314953520704 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 13:57:43.477144 123422613400704 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 13:57:43.477252 123422613400704 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.228444 seconds. If this number is high, consider checkpointing less frequently. I0417 13:57:43.477465 123422613400704 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0417 13:57:43.477770 123422613400704 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_false/checkpoints/19. I0417 13:57:43.644903 123422613400704 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:57:43.644996 123422613400704 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 13:57:43.652318 123314953520704 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19 I0417 13:57:44.268692 123314974492224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19/items I0417 13:57:51.274093 123422613400704 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.629569s I0417 13:57:51.278718 123422613400704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.599 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0417 13:57:51.278797 123422613400704 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.717131s (batch_requests_ready=0.079947s, total_serialization_initiated=7.632699s, others=0.004485s) I0417 13:57:51.278889 123422613400704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.717882s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=7.717871) I0417 13:57:51.280388 123314953520704 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:57:51.280468 123422613400704 async_checkpointer.py:561] Finished blocking save. Time taken: 7.802963s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19. I0417 13:57:51.280671 123422613400704 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:57:51.280886 123313936401984 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:57:51.280985 123422613400704 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434256.2487936, 'wait_for_prev_duration_secs': 7.228443622589111, 'time_between_consecutive_saves_sec': 5.85493803024292, 'checkpointer_blocking_start_time': 1776434263.4774873, 'checkpointer_blocking_duration_secs': 7.803089141845703, 'get_old_steps_start_time': 1776434271.2805977, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776434256.248739, 'checkpoint_manager_blocking_duration_secs': 15.03221869468689} I0417 13:57:51.281147 123422613400704 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0417 13:57:51.281181 123422613400704 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:57:52.033212 123314964006464 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_false/checkpoints/19/items/array_metadatas/process_0 I0417 13:57:58.165441 123314974492224 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.885922s (commit=6.479882s, array_metadata_write=0.406040s) I0417 13:57:58.166347 123314953520704 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 865.258 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 14 seconds) (per-host) I0417 13:57:58.166397 123314953520704 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.885886s. W0417 13:57:58.475921 123314953520704 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19/items, I0417 13:57:58.571842 123314953520704 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 13:57:58.571956 123314953520704 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.139018s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19/items I0417 13:57:58.572611 123314953520704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19/items I0417 13:57:58.808806 123314953520704 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19 I0417 13:57:59.483372 123314953520704 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_false/checkpoints/19`. I0417 13:57:59.484554 123314953520704 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 16.007055s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_15_ocdbt_false/checkpoints/19 I0417 13:57:59.484619 123314953520704 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 8.204109s. I0417 13:57:59.484778 123313936401984 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:57:59.484875 123313936401984 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:57:59.484926 123313936401984 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:57:59.484964 123313936401984 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 13:57:59.485113 123422613400704 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 13:57:59.485249 123422613400704 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:57:59.486115 123422613400704 metric_logger.py:196] completed step: 19, seconds: 0.122, TFLOP/s/device: 111.600, Tokens/s/device: 16821.493, 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