2026-04-17 13:14:47.415633: 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 13:14:47.535871 126551392418944 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:15:33.077257 126551392418944 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:15:33.077370 126551392418944 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:15:33.077405 126551392418944 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 13:15:33.077429 126551392418944 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:15:33.077451 126551392418944 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:15:33.077569 126551392418944 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:15:33.169517 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:33.570000 126551392418944 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:15:33.570322 126551392418944 checkpointing.py:677] Setting up checkpoint logger... I0417 13:15:33.570370 126551392418944 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:15:33.570417 126551392418944 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:15:33.570987 126551392418944 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 0x7318741f8a40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:15:35.208637 126551392418944 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:15:35.208825 126551392418944 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 0x7311f8332210>}, handler_registry=None I0417 13:15:35.209131 126551392418944 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7311f8332210>` 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 13:15:35.209175 126551392418944 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7311f8331010>` 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 13:15:35.209203 126551392418944 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 0x7311f8332210>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7311f8332210>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7311f8331010>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7311f8331010>}). I0417 13:15:35.209491 126551392418944 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:15:35.209541 126551392418944 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7311f832e160> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:15:35.600489 126445292619328 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431538118742465, 'commit_timestamp_nsecs': 1776431566259917925, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:15:35.687982 126445292619328 checkpoint_manager.py:1847] Missing metrics for step 0 I0417 13:15:35.725203 126551392418944 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints I0417 13:15:35.725678 126551392418944 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_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7311f8332450> I0417 13:15:35.725780 126551392418944 checkpointing.py:301] Checkpoint manager created! W0417 13:15:35.742562 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:36.262552 126551392418944 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:15:36.262656 126551392418944 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:15:36.345491 126551392418944 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:15:36.345567 126551392418944 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:15:36.359935 126551392418944 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:15:36.359983 126551392418944 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:15:36.380520 126551392418944 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 13:15:36.380573 126551392418944 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:15:36.394955 126551392418944 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 13:15:36.395003 126551392418944 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:15:36.409296 126551392418944 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 13:15:36.409343 126551392418944 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:15:36.430954 126551392418944 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:15:36.431009 126551392418944 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:15:36.449360 126551392418944 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:15:36.449412 126551392418944 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:15:36.639201 126551392418944 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:15:36.639291 126551392418944 checkpointing.py:581] restoring from this run's directory step 9 I0417 13:15:36.797983 126551392418944 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/9. I0417 13:15:37.323390 445997 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:15:37.323457 445997 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:15:42.830768 126551392418944 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 2.163 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0417 13:15:42.831084 126551392418944 checkpointer.py:318] Finished restoring checkpoint in 6.11 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/9. I0417 13:15:42.831171 126551392418944 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776431736.7204766, 'checkpointer_duration_secs': 6.110656976699829, 'checkpoint_manager_start_time': 1776431736.6394937, 'checkpoint_manager_duration_secs': 6.191643714904785} [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 13:15:43.215255 126551392418944 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:15:43.215359 126551392418944 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 13:15:43.215400 126551392418944 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 13:15:43.215444 126551392418944 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 13:15:43.215484 126551392418944 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 13:15:43.215511 126551392418944 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 13:15:43.215549 126551392418944 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 13:15:43.215586 126551392418944 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 13:15:43.215616 126551392418944 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 13:15:43.215642 126551392418944 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 13:15:43.215677 126551392418944 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:15:43.215711 126551392418944 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:15:43.634586 126551392418944 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:15:43.634666 126551392418944 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:15:43.649468 126551392418944 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:15:43.649519 126551392418944 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:15:43.875928 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:43.939240 126551392418944 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 13:15:43.939785 126551392418944 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:15:43.939951 126551392418944 metric_logger.py:301] number parameters: 1.104 billion W0417 13:15:44.753190 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:44.818813 126551392418944 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0417 13:15:44.934981 126551392418944 checkpointing.py:776] Waited 0.11615514755249023 seconds for step 10 to finish before starting checkpointing. I0417 13:15:44.935417 126551392418944 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:15:44.935578 126551392418944 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0417 13:15:44.935878 126551392418944 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10. I0417 13:15:45.015317 126551392418944 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:15:45.088992 126443700880960 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10 I0417 13:15:45.110575 126551392418944 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:15:45.110660 126551392418944 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 13:15:45.327273 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.355255 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.360713 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.365328 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.369610 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.373688 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.377688 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:45.381650 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:45.744858 126443690395200 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10/items I0417 13:15:45.931428 126443658937920 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431745667093470, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776431745667093470, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0417 13:15:52.415707 126551392418944 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.305553s I0417 13:15:52.419906 126551392418944 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.667 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0417 13:15:52.419981 126551392418944 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.403398s (batch_requests_ready=0.090960s, total_serialization_initiated=7.308331s, others=0.004107s) I0417 13:15:52.420072 126551392418944 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.404313s (all_items=0.000031s, per_item={'items': '0.00003147'}, temp_paths=7.404282) I0417 13:15:52.421660 126443679909440 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:15:52.421745 126551392418944 async_checkpointer.py:561] Finished blocking save. Time taken: 7.486125s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10. I0417 13:15:52.421935 126551392418944 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:15:52.422126 126443711366720 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:15:52.422229 126551392418944 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776431744.935403, 'wait_for_prev_duration_secs': 4.00543212890625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776431744.9355986, 'checkpointer_blocking_duration_secs': 7.486251354217529, 'get_old_steps_start_time': 1776431752.4218664, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776431744.9353044, 'checkpoint_manager_blocking_duration_secs': 7.486897945404053} I0417 13:15:52.422338 126551392418944 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0417 13:15:52.422378 126551392418944 max_utils.py:750] Memstats: After params initialized: I0417 13:15:52.422440 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:15:52.422466 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:15:52.422486 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:15:52.422506 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:15:52.422525 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:15:52.422543 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:15:52.422560 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:15:52.422576 126551392418944 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:15:52.428458 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:52.434788 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:15:52.439665 443728 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:15:52.623899 126551392418944 metric_logger.py:196] completed step: 10, seconds: 0.878, TFLOP/s/device: 15.469, Tokens/s/device: 2331.703, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0417 13:15:52.740283 126551392418944 metric_logger.py:196] completed step: 11, seconds: 7.803, TFLOP/s/device: 1.741, Tokens/s/device: 262.452, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0417 13:15:52.858372 126551392418944 metric_logger.py:196] completed step: 12, seconds: 0.005, TFLOP/s/device: 2581.632, Tokens/s/device: 389131.674, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0417 13:15:52.976331 126551392418944 metric_logger.py:196] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.728, Tokens/s/device: 17443.742, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0417 13:15:52.980684 126551392418944 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0417 13:15:53.126529 126443669423680 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_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_0 I0417 13:15:53.212119 126551392418944 checkpointing.py:776] Waited 0.2313988208770752 seconds for step 15 to finish before starting checkpointing. I0417 13:15:53.212508 126551392418944 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:16:30.280138 126443690395200 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.859349s (commit=37.416735s, array_metadata_write=0.442614s) I0417 13:16:30.281235 126443679909440 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 279.176 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0417 13:16:30.281295 126443679909440 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.859508s. I0417 13:16:30.738415 126443679909440 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 13:16:30.912538 126443658937920 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776431745667093470, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA I0417 13:16:31.122764 126443679909440 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:16:31.123323 126443679909440 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.545785s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10/items I0417 13:16:31.123967 126443679909440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10/items I0417 13:16:31.361216 126443679909440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10 I0417 13:16:32.021469 126443679909440 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_08_checkpoint_async_true/checkpoints/10`. I0417 13:16:32.022083 126443679909440 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.086471s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/10 I0417 13:16:32.022159 126443679909440 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.600373s. I0417 13:16:32.022318 126443711366720 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:16:32.022420 126443711366720 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:16:32.022497 126443711366720 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:16:32.022550 126443711366720 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 13:16:32.023170 126551392418944 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 13:16:32.023301 126551392418944 checkpoint_manager.py:1441] Waiting for previous save to complete took 38.810790 seconds. If this number is high, consider checkpointing less frequently. I0417 13:16:32.024101 126551392418944 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0417 13:16:32.024392 126551392418944 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15. I0417 13:16:32.191249 126551392418944 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:16:32.191339 126551392418944 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 13:16:32.206164 126443711366720 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15 I0417 13:16:32.885446 126443648452160 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15/items I0417 13:16:40.794274 126551392418944 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.603426s I0417 13:16:40.800108 126551392418944 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.420 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0417 13:16:40.800194 126551392418944 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.693721s (batch_requests_ready=0.081214s, total_serialization_initiated=8.606833s, others=0.005674s) I0417 13:16:40.800291 126551392418944 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.694273s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=8.694257) I0417 13:16:40.802346 126443711366720 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:16:40.802494 126551392418944 async_checkpointer.py:561] Finished blocking save. Time taken: 8.778339s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15. I0417 13:16:40.802784 126551392418944 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:16:40.802975 126443679909440 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:16:40.803109 126551392418944 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776431753.2124846, 'wait_for_prev_duration_secs': 38.810789585113525, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776431792.0241258, 'checkpointer_blocking_duration_secs': 8.778519630432129, 'get_old_steps_start_time': 1776431800.802677, 'get_old_steps_duration_secs': 4.0531158447265625e-05, 'checkpoint_manager_blocking_start_time': 1776431753.2124205, 'checkpoint_manager_blocking_duration_secs': 47.590651750564575} I0417 13:16:40.803240 126551392418944 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0417 13:16:40.803994 126551392418944 metric_logger.py:196] completed step: 14, seconds: 0.118, TFLOP/s/device: 115.196, Tokens/s/device: 17363.584, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0417 13:16:40.813635 126551392418944 metric_logger.py:196] completed step: 15, seconds: 0.118, TFLOP/s/device: 115.348, Tokens/s/device: 17386.432, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0417 13:16:40.928774 126551392418944 metric_logger.py:196] completed step: 16, seconds: 47.830, TFLOP/s/device: 0.284, Tokens/s/device: 42.818, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0417 13:16:41.046898 126551392418944 metric_logger.py:196] completed step: 17, seconds: 0.007, TFLOP/s/device: 1966.015, Tokens/s/device: 296339.169, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0417 13:16:41.165934 126551392418944 metric_logger.py:196] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.303, Tokens/s/device: 17530.494, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0417 13:16:41.283496 126551392418944 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0417 13:16:41.284845 126551392418944 checkpointing.py:776] Waited 0.0013911724090576172 seconds for step 19 to finish before starting checkpointing. I0417 13:16:41.285209 126551392418944 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:16:41.566941 126445240190528 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_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_0 I0417 13:17:16.724450 126443648452160 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.923514s (commit=35.509835s, array_metadata_write=0.413679s) I0417 13:17:16.726189 126443711366720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 283.212 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0417 13:17:16.726362 126443711366720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.923807s. I0417 13:17:17.164104 126443711366720 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 13:17:17.582201 126443711366720 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:17:17.583279 126443711366720 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.564755s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15/items I0417 13:17:17.584214 126443711366720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15/items I0417 13:17:17.834026 126443711366720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15 I0417 13:17:18.478606 126443711366720 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_08_checkpoint_async_true/checkpoints/15`. I0417 13:17:18.479513 126443711366720 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.455369s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/15 I0417 13:17:18.479610 126443711366720 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.677060s. I0417 13:17:18.479797 126443679909440 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:17:18.479866 126443679909440 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:17:18.479978 126443679909440 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:17:18.480036 126443679909440 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 13:17:18.480827 126551392418944 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 13:17:18.481270 126551392418944 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.196076 seconds. If this number is high, consider checkpointing less frequently. I0417 13:17:18.481541 126551392418944 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0417 13:17:18.481884 126551392418944 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19. I0417 13:17:18.671294 126551392418944 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:17:18.671382 126551392418944 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 13:17:18.675494 126443679909440 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19 I0417 13:17:19.377063 126445229704768 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19/items I0417 13:17:25.215348 126551392418944 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.544445s I0417 13:17:25.220091 126551392418944 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.859 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:17:25.220162 126551392418944 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.639589s (batch_requests_ready=0.087062s, total_serialization_initiated=6.547888s, others=0.004639s) I0417 13:17:25.220242 126551392418944 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.640386s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=6.640370) I0417 13:17:25.221805 126443711366720 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:17:25.221907 126551392418944 async_checkpointer.py:561] Finished blocking save. Time taken: 6.740318s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19. I0417 13:17:25.222106 126551392418944 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:17:25.222291 126443679909440 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:17:25.222401 126551392418944 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776431801.2851722, 'wait_for_prev_duration_secs': 37.196075677871704, 'time_between_consecutive_saves_sec': 9.262607336044312, 'checkpointer_blocking_start_time': 1776431838.4815674, 'checkpointer_blocking_duration_secs': 6.740452289581299, 'get_old_steps_start_time': 1776431845.222036, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776431801.285125, 'checkpoint_manager_blocking_duration_secs': 43.937252044677734} I0417 13:17:25.222634 126551392418944 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0417 13:17:25.222670 126551392418944 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:17:25.936006 126445240190528 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_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_0 I0417 13:17:58.942594 126445229704768 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.721718s (commit=33.254704s, array_metadata_write=0.467014s) I0417 13:17:58.943623 126443711366720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 313.078 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0417 13:17:58.943737 126443711366720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.721779s. I0417 13:17:59.340166 126443711366720 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 13:17:59.760482 126443711366720 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:17:59.760653 126443711366720 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.563958s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19/items I0417 13:17:59.761286 126443711366720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19/items I0417 13:17:59.993758 126443711366720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19 I0417 13:18:00.667889 126443711366720 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_08_checkpoint_async_true/checkpoints/19`. I0417 13:18:00.668500 126443711366720 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.186917s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_08_checkpoint_async_true/checkpoints/19 I0417 13:18:00.668566 126443711366720 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.446614s. I0417 13:18:00.668730 126443679909440 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:18:00.668831 126443679909440 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:18:00.668903 126443679909440 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:18:00.668954 126443679909440 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 13:18:00.669109 126551392418944 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 13:18:00.669252 126551392418944 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:18:00.669969 126551392418944 metric_logger.py:196] completed step: 19, seconds: 0.118, TFLOP/s/device: 115.073, Tokens/s/device: 17345.055, 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