2026-04-22 18:48:56.081249: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776883736.094298 2082059 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776883736.098145 2082059 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776883736.109647 2082059 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883736.109661 2082059 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883736.109663 2082059 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883736.109664 2082059 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-22 18:48:59.767921: 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) I0422 18:48:59.902975 128153810356032 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 18:49:30.984935 128153810356032 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:49:30.985230 128153810356032 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:49:30.985282 128153810356032 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0422 18:49:30.985311 128153810356032 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 18:49:30.985336 128153810356032 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:49:30.985506 128153810356032 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 18:49:31.637968 128153810356032 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:49:31.638255 128153810356032 checkpointing.py:677] Setting up checkpoint logger... I0422 18:49:31.638300 128153810356032 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 18:49:31.638336 128153810356032 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 18:49:31.638752 128153810356032 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 0x748ddb5adc70>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:49:33.363365 128153810356032 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 18:49:33.363754 128153810356032 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 0x748df2f642c0>}, handler_registry=None I0422 18:49:33.363985 128153810356032 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x748df2f642c0>` 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`. I0422 18:49:33.364028 128153810356032 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748c04fb5850>` 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`. I0422 18:49:33.364072 128153810356032 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 0x748df2f642c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x748df2f642c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748c04fb5850>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748c04fb5850>}). I0422 18:49:33.364346 128153810356032 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 18:49:33.364398 128153810356032 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x748c04fc4900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:49:33.818595 128124064892480 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883668968501130, 'commit_timestamp_nsecs': 1776883713453932228, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0422 18:49:33.901297 128124075378240 checkpoint_manager.py:1847] Missing metrics for step 5 I0422 18:49:33.957905 128153810356032 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints I0422 18:49:33.958457 128153810356032 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_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x748c04fb61e0> I0422 18:49:33.958567 128153810356032 checkpointing.py:301] Checkpoint manager created! I0422 18:49:34.364339 128153810356032 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:49:34.364477 128153810356032 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:49:34.649360 128153810356032 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:49:34.649452 128153810356032 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:49:34.661703 128153810356032 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:49:34.661759 128153810356032 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:49:34.680808 128153810356032 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:49:34.680878 128153810356032 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:49:34.693262 128153810356032 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:49:34.693314 128153810356032 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:49:34.705644 128153810356032 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:49:34.705698 128153810356032 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:49:34.725213 128153810356032 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:49:34.725272 128153810356032 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:49:34.740792 128153810356032 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:49:34.740847 128153810356032 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 18:49:34.891817 128153810356032 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:49:34.891923 128153810356032 checkpointing.py:581] restoring from this run's directory step 9 I0422 18:49:35.089589 128153810356032 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9. I0422 18:49:35.638571 2083632 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 18:49:35.638644 2083632 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 18:49:42.997775 128153810356032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.630 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0422 18:49:42.998174 128153810356032 checkpointer.py:318] Finished restoring checkpoint in 8.03 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9. I0422 18:49:42.998257 128153810356032 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776883774.9723892, 'checkpointer_duration_secs': 8.025840044021606, 'checkpoint_manager_start_time': 1776883774.892169, 'checkpoint_manager_duration_secs': 8.106061935424805} [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 I0422 18:49:43.379434 128153810356032 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 18:49:43.379548 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 18:49:43.379590 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 18:49:43.379635 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0422 18:49:43.379674 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 18:49:43.379703 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 18:49:43.379749 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 18:49:43.379790 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0422 18:49:43.379820 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0422 18:49:43.379848 128153810356032 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 18:49:43.379885 128153810356032 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 18:49:43.379921 128153810356032 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 18:49:43.782686 128153810356032 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:49:43.782776 128153810356032 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:49:43.794716 128153810356032 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:49:43.794771 128153810356032 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:49:44.567098 128153810356032 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. I0422 18:49:44.570944 128153810356032 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 18:49:44.571244 128153810356032 metric_logger.py:301] number parameters: 1.104 billion I0422 18:49:46.562161 128153810356032 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0422 18:49:46.679360 128153810356032 checkpointing.py:776] Waited 0.11719989776611328 seconds for step 10 to finish before starting checkpointing. I0422 18:49:46.680066 128153810356032 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:49:46.680263 128153810356032 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0422 18:49:46.680671 128153810356032 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10. I0422 18:49:46.773316 128153810356032 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 18:49:46.856448 128153810356032 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:49:46.856552 128153810356032 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 I0422 18:49:46.862988 128121881757248 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10 I0422 18:49:47.599629 128121871271488 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/items I0422 18:49:47.770669 128121839814208 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883787520751535, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776883787520751535, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0422 18:49:49.565540 128153810356032 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.709766s I0422 18:49:49.570058 128153810356032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.415 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 18:49:49.570129 128153810356032 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.795411s (batch_requests_ready=0.078309s, total_serialization_initiated=2.712690s, others=0.004412s) I0422 18:49:49.570209 128153810356032 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.796346s (all_items=0.000025s, per_item={'items': '0.00002527'}, temp_paths=2.796321) I0422 18:49:49.571657 128121892243008 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:49:49.571803 128153810356032 async_checkpointer.py:561] Finished blocking save. Time taken: 2.891495s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10. I0422 18:49:49.572020 128153810356032 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:49:49.572257 128121860785728 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:49:49.572371 128153810356032 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883786.6800346, 'wait_for_prev_duration_secs': 6.222724914550781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883786.6802864, 'checkpointer_blocking_duration_secs': 2.8916447162628174, 'get_old_steps_start_time': 1776883789.5719464, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776883786.6799097, 'checkpoint_manager_blocking_duration_secs': 2.892435073852539} I0422 18:49:49.572453 128153810356032 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0422 18:49:49.572499 128153810356032 max_utils.py:750] Memstats: After params initialized: I0422 18:49:49.572548 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 18:49:49.572572 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 18:49:49.572594 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 18:49:49.572613 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 18:49:49.572631 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 18:49:49.572648 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 18:49:49.572665 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 18:49:49.572681 128153810356032 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 18:49:49.814777 128153810356032 metric_logger.py:196] completed step: 10, seconds: 1.990, TFLOP/s/device: 6.827, Tokens/s/device: 1029.105, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0422 18:49:49.931506 128153810356032 metric_logger.py:196] completed step: 11, seconds: 3.251, TFLOP/s/device: 4.180, Tokens/s/device: 630.045, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0422 18:49:50.050131 128153810356032 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2161.490, Tokens/s/device: 325803.373, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0422 18:49:50.169467 128153810356032 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 115.510, Tokens/s/device: 17410.969, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0422 18:49:50.175440 128153810356032 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0422 18:49:50.330028 128121850299968 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0422 18:49:50.407222 128153810356032 checkpointing.py:776] Waited 0.23174738883972168 seconds for step 15 to finish before starting checkpointing. I0422 18:49:50.408236 128153810356032 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:50:37.621520 128121871271488 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 48.050622s (commit=47.548046s, array_metadata_write=0.502576s) I0422 18:50:37.630903 128121892243008 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 248.481 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0422 18:50:37.631014 128121892243008 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 48.059167s. I0422 18:50:38.180399 128121892243008 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 18:50:38.393827 128121839814208 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883787520751535, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0422 18:50:38.567557 128121892243008 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:50:38.567738 128121892243008 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.574647s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/items I0422 18:50:38.568500 128121892243008 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10/items I0422 18:50:38.816831 128121892243008 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10 I0422 18:50:39.519789 128121892243008 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10`. I0422 18:50:39.520522 128121892243008 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.840221s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/10 I0422 18:50:39.520595 128121892243008 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 49.948749s. I0422 18:50:39.520820 128121860785728 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:50:39.520947 128121860785728 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:50:39.521020 128121860785728 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:50:39.521083 128121860785728 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 18:50:39.521255 128153810356032 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. W0422 18:50:39.521407 128153810356032 checkpoint_manager.py:1441] Waiting for previous save to complete took 49.113163 seconds. If this number is high, consider checkpointing less frequently. I0422 18:50:39.522418 128153810356032 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0422 18:50:39.522734 128153810356032 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15. I0422 18:50:39.692272 128153810356032 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:50:39.692369 128153810356032 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 I0422 18:50:39.695266 128121892243008 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15 I0422 18:50:40.432754 128121881757248 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15/items I0422 18:50:50.106840 128153810356032 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.415028s I0422 18:50:50.111288 128153810356032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.174 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0422 18:50:50.111359 128153810356032 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.509821s (batch_requests_ready=0.086779s, total_serialization_initiated=10.418802s, others=0.004240s) I0422 18:50:50.111461 128153810356032 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.510427s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=10.510411) I0422 18:50:50.113710 128121860785728 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:50:50.113801 128153810356032 async_checkpointer.py:561] Finished blocking save. Time taken: 10.591339s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15. I0422 18:50:50.114024 128153810356032 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:50:50.114249 128121871271488 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:50:50.114372 128153810356032 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883790.4082112, 'wait_for_prev_duration_secs': 49.113163232803345, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883839.5224433, 'checkpointer_blocking_duration_secs': 10.59147572517395, 'get_old_steps_start_time': 1776883850.1139412, 'get_old_steps_duration_secs': 3.552436828613281e-05, 'checkpoint_manager_blocking_start_time': 1776883790.4081442, 'checkpoint_manager_blocking_duration_secs': 59.706199169158936} I0422 18:50:50.114530 128153810356032 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0422 18:50:50.115237 128153810356032 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.301, Tokens/s/device: 17228.616, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0422 18:50:50.124555 128153810356032 metric_logger.py:196] completed step: 15, seconds: 0.120, TFLOP/s/device: 113.489, Tokens/s/device: 17106.296, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0422 18:50:50.240411 128153810356032 metric_logger.py:196] completed step: 16, seconds: 59.947, TFLOP/s/device: 0.227, Tokens/s/device: 34.164, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0422 18:50:50.358834 128153810356032 metric_logger.py:196] completed step: 17, seconds: 0.007, TFLOP/s/device: 1951.333, Tokens/s/device: 294126.095, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0422 18:50:50.477955 128153810356032 metric_logger.py:196] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.197, Tokens/s/device: 17514.453, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0422 18:50:50.596818 128153810356032 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0422 18:50:50.598200 128153810356032 checkpointing.py:776] Waited 0.0014064311981201172 seconds for step 19 to finish before starting checkpointing. I0422 18:50:50.598561 128153810356032 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:50:50.863946 128123905508928 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0422 18:51:33.033612 128123895023168 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.921413s (commit=42.382828s, array_metadata_write=0.538585s) I0422 18:51:33.034726 128121860785728 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 236.497 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 53 seconds) (per-host) I0422 18:51:33.034846 128121860785728 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.921000s. I0422 18:51:33.529625 128121860785728 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 18:51:33.972888 128121860785728 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:51:33.973129 128121860785728 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.603741s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15/items I0422 18:51:33.974012 128121860785728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15/items I0422 18:51:34.220555 128121860785728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15 I0422 18:51:34.946496 128121860785728 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15`. I0422 18:51:34.947216 128121860785728 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 55.424761s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/15 I0422 18:51:34.947291 128121860785728 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 44.833448s. I0422 18:51:34.947489 128121871271488 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:51:34.947610 128121871271488 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:51:34.947672 128121871271488 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:51:34.947715 128121871271488 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 18:51:34.954164 128153810356032 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. W0422 18:51:34.954287 128153810356032 checkpoint_manager.py:1441] Waiting for previous save to complete took 44.355744 seconds. If this number is high, consider checkpointing less frequently. I0422 18:51:34.954570 128153810356032 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0422 18:51:34.954877 128153810356032 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19. I0422 18:51:35.138557 128121860785728 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19 I0422 18:51:35.148381 128153810356032 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:51:35.148483 128153810356032 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 I0422 18:51:35.833534 128123905508928 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19/items I0422 18:51:37.715918 128153810356032 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.567955s I0422 18:51:37.719765 128153810356032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.630 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 18:51:37.719827 128153810356032 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.665195s (batch_requests_ready=0.090307s, total_serialization_initiated=2.571174s, others=0.003714s) I0422 18:51:37.719897 128153810356032 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.665757s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=2.665740) I0422 18:51:37.721787 128121871271488 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:51:37.721871 128153810356032 async_checkpointer.py:561] Finished blocking save. Time taken: 2.767261s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19. I0422 18:51:37.722034 128153810356032 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:51:37.722233 128123895023168 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:51:37.722348 128153810356032 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883850.5985286, 'wait_for_prev_duration_secs': 44.35574436187744, 'time_between_consecutive_saves_sec': 11.077428817749023, 'checkpointer_blocking_start_time': 1776883894.954594, 'checkpointer_blocking_duration_secs': 2.767378568649292, 'get_old_steps_start_time': 1776883897.721982, 'get_old_steps_duration_secs': 1.9311904907226562e-05, 'checkpoint_manager_blocking_start_time': 1776883850.5984738, 'checkpoint_manager_blocking_duration_secs': 47.12385058403015} I0422 18:51:37.722431 128153810356032 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0422 18:51:37.722461 128153810356032 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:51:38.472688 128123863565888 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0422 18:52:08.095218 128123905508928 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.374543s (commit=29.854585s, array_metadata_write=0.519957s) I0422 18:52:08.096532 128121871271488 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 382.448 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0422 18:52:08.096661 128121871271488 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.374745s. I0422 18:52:08.630221 128121871271488 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 18:52:09.042017 128121871271488 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:52:09.042223 128121871271488 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.608565s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19/items I0422 18:52:09.042962 128121871271488 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19/items I0422 18:52:09.279522 128121871271488 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19 I0422 18:52:10.016344 128121871271488 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19`. I0422 18:52:10.017090 128121871271488 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.062485s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/19 I0422 18:52:10.017155 128121871271488 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.295242s. I0422 18:52:10.017372 128123895023168 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:52:10.017502 128123895023168 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:52:10.017562 128123895023168 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:52:10.017641 128123895023168 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 18:52:10.017740 128153810356032 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. I0422 18:52:10.017839 128153810356032 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:52:10.019096 128153810356032 metric_logger.py:196] completed step: 19, seconds: 0.118, TFLOP/s/device: 114.738, Tokens/s/device: 17294.522, 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