2026-04-22 19:00:20.863924: 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 00:00:1776884420.876938 2093833 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776884420.880774 2093833 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776884420.892202 2093833 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884420.892214 2093833 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884420.892216 2093833 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884420.892217 2093833 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 19:00:24.552643: 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 19:00:24.687462 125098811889472 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 19:00:54.846937 125098811889472 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 19:00:54.847122 125098811889472 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 19:00:54.847159 125098811889472 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 19:00:54.847191 125098811889472 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 19:00:54.847216 125098811889472 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 19:00:54.847332 125098811889472 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 19:00:55.492955 125098811889472 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 19:00:55.493316 125098811889472 checkpointing.py:677] Setting up checkpoint logger... I0422 19:00:55.493362 125098811889472 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 19:00:55.493434 125098811889472 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 19:00:55.494020 125098811889472 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 0x71c68cbb5d00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 19:00:57.170856 125098811889472 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 19:00:57.171315 125098811889472 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 0x71c4b3fa1970>}, handler_registry=None I0422 19:00:57.171649 125098811889472 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71c4b3fa1970>` 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 19:00:57.171695 125098811889472 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71c4b3fa1f40>` 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 19:00:57.171725 125098811889472 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 0x71c4b3fa1970>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x71c4b3fa1970>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71c4b3fa1f40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x71c4b3fa1f40>}). I0422 19:00:57.172027 125098811889472 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 19:00:57.172111 125098811889472 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x71c4b3fb8900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 19:00:57.625709 125069078562368 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884273318634917, 'commit_timestamp_nsecs': 1776884312963224910, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0422 19:00:57.715682 125069057590848 checkpoint_manager.py:1847] Missing metrics for step 5 I0422 19:00:57.749575 125098811889472 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints I0422 19:00:57.750179 125098811889472 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_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x71c4b3fa1e50> I0422 19:00:57.750299 125098811889472 checkpointing.py:301] Checkpoint manager created! I0422 19:00:58.131069 125098811889472 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 19:00:58.131170 125098811889472 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:00:58.416412 125098811889472 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 19:00:58.416501 125098811889472 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:00:58.428669 125098811889472 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 19:00:58.428720 125098811889472 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:00:58.447797 125098811889472 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 19:00:58.447863 125098811889472 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:00:58.460164 125098811889472 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 19:00:58.460216 125098811889472 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:00:58.472440 125098811889472 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 19:00:58.472491 125098811889472 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:00:58.491810 125098811889472 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 19:00:58.491872 125098811889472 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:00:58.507280 125098811889472 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 19:00:58.507335 125098811889472 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 19:00:58.657561 125098811889472 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 19:00:58.657679 125098811889472 checkpointing.py:581] restoring from this run's directory step 9 I0422 19:00:58.836737 125098811889472 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/9. I0422 19:00:59.421205 2095371 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 19:00:59.421277 2095371 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 19:01:06.730550 125098811889472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.635 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0422 19:01:06.730914 125098811889472 checkpointer.py:318] Finished restoring checkpoint in 7.98 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/9. I0422 19:01:06.731021 125098811889472 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776884458.7532992, 'checkpointer_duration_secs': 7.9776740074157715, 'checkpoint_manager_start_time': 1776884458.6579154, 'checkpoint_manager_duration_secs': 8.073059320449829} [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 19:01:07.055821 125098811889472 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 19:01:07.055937 125098811889472 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 19:01:07.055980 125098811889472 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 19:01:07.056026 125098811889472 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 19:01:07.056085 125098811889472 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 19:01:07.056117 125098811889472 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 19:01:07.056158 125098811889472 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 19:01:07.056200 125098811889472 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 19:01:07.056230 125098811889472 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 19:01:07.056257 125098811889472 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 19:01:07.056294 125098811889472 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 19:01:07.056330 125098811889472 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 19:01:07.458973 125098811889472 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 19:01:07.459076 125098811889472 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 19:01:07.470671 125098811889472 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 19:01:07.470722 125098811889472 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 19:01:08.235930 125098811889472 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 19:01:08.239462 125098811889472 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 19:01:08.239625 125098811889472 metric_logger.py:301] number parameters: 1.104 billion I0422 19:01:10.003078 125098811889472 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0422 19:01:10.120755 125098811889472 checkpointing.py:776] Waited 0.11767125129699707 seconds for step 10 to finish before starting checkpointing. I0422 19:01:10.121414 125098811889472 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 19:01:10.121568 125098811889472 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0422 19:01:10.121874 125098811889472 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10. I0422 19:01:10.213850 125098811889472 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 19:01:10.310009 125098811889472 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:01:10.310128 125098811889472 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 19:01:10.324425 125066853484096 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10 I0422 19:01:11.056413 125066842998336 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10/items I0422 19:01:11.233967 125066446636608 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884470967859601, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776884470967859601, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0422 19:01:13.219722 125098811889472 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.910135s I0422 19:01:13.223948 125098811889472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.101 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 19:01:13.224011 125098811889472 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.009112s (batch_requests_ready=0.091965s, total_serialization_initiated=2.913040s, others=0.004107s) I0422 19:01:13.224108 125098811889472 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.009703s (all_items=0.000021s, per_item={'items': '0.00002074'}, temp_paths=3.009682) I0422 19:01:13.226073 125066863969856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:01:13.226161 125098811889472 async_checkpointer.py:561] Finished blocking save. Time taken: 3.104553s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10. I0422 19:01:13.226338 125098811889472 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:01:13.226545 125066255795776 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:01:13.226657 125098811889472 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884470.1214006, 'wait_for_prev_duration_secs': 4.458427429199219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884470.1215894, 'checkpointer_blocking_duration_secs': 3.104677200317383, 'get_old_steps_start_time': 1776884473.2262797, 'get_old_steps_duration_secs': 2.288818359375e-05, 'checkpoint_manager_blocking_start_time': 1776884470.1213162, 'checkpoint_manager_blocking_duration_secs': 3.105314016342163} I0422 19:01:13.226784 125098811889472 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0422 19:01:13.226823 125098811889472 max_utils.py:750] Memstats: After params initialized: I0422 19:01:13.226873 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 19:01:13.226898 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 19:01:13.226917 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 19:01:13.226936 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 19:01:13.226953 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 19:01:13.226970 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 19:01:13.226987 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 19:01:13.227004 125098811889472 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 19:01:13.479837 125098811889472 metric_logger.py:196] completed step: 10, seconds: 1.763, TFLOP/s/device: 7.707, Tokens/s/device: 1161.743, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0422 19:01:13.596583 125098811889472 metric_logger.py:196] completed step: 11, seconds: 3.475, TFLOP/s/device: 3.910, Tokens/s/device: 589.401, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0422 19:01:13.715925 125098811889472 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2132.318, Tokens/s/device: 321406.152, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0422 19:01:13.834792 125098811889472 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 115.327, Tokens/s/device: 17383.333, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0422 19:01:13.840327 125098811889472 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0422 19:01:13.994601 125066492773952 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_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0 I0422 19:01:14.072666 125098811889472 checkpointing.py:776] Waited 0.2322986125946045 seconds for step 15 to finish before starting checkpointing. I0422 19:01:14.073677 125098811889472 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:02:01.457664 125066842998336 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 48.232693s (commit=47.742229s, array_metadata_write=0.490464s) I0422 19:02:01.458920 125066863969856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 246.600 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 51 seconds) (per-host) I0422 19:02:01.459056 125066863969856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 48.232840s. I0422 19:02:01.946887 125066863969856 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 19:02:02.156368 125066446636608 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884470967859601, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0422 19:02:02.459034 125066863969856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 19:02:02.459299 125066863969856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.688993s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10/items I0422 19:02:02.460210 125066863969856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10/items I0422 19:02:02.734517 125066863969856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10 I0422 19:02:03.487380 125066863969856 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_16_zarr3_true/checkpoints/10`. I0422 19:02:03.488076 125066863969856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 53.366474s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/10 I0422 19:02:03.488158 125066863969856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 50.261955s. I0422 19:02:03.488369 125066255795776 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:02:03.488488 125066255795776 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:02:03.488560 125066255795776 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:02:03.488613 125066255795776 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 19:02:03.488788 125098811889472 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 19:02:03.488921 125098811889472 checkpoint_manager.py:1441] Waiting for previous save to complete took 49.415241 seconds. If this number is high, consider checkpointing less frequently. I0422 19:02:03.490316 125098811889472 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0422 19:02:03.490627 125098811889472 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15. I0422 19:02:03.667155 125098811889472 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:02:03.667252 125098811889472 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 19:02:03.677409 125066863969856 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15 I0422 19:02:04.389257 125066492773952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15/items I0422 19:02:07.459494 125098811889472 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.792774s I0422 19:02:07.463742 125098811889472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.181 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 19:02:07.463807 125098811889472 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.880185s (batch_requests_ready=0.079790s, total_serialization_initiated=3.796256s, others=0.004139s) I0422 19:02:07.463881 125098811889472 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.880762s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=3.880748) I0422 19:02:07.465558 125066842998336 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:02:07.465665 125098811889472 async_checkpointer.py:561] Finished blocking save. Time taken: 3.975303s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15. I0422 19:02:07.465869 125098811889472 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:02:07.466143 125066255795776 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:02:07.466262 125098811889472 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884474.0736527, 'wait_for_prev_duration_secs': 49.41524124145508, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884523.4903421, 'checkpointer_blocking_duration_secs': 3.9754371643066406, 'get_old_steps_start_time': 1776884527.4657974, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776884474.0735903, 'checkpoint_manager_blocking_duration_secs': 53.3926465511322} I0422 19:02:07.466422 125098811889472 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0422 19:02:07.467114 125098811889472 metric_logger.py:196] completed step: 14, seconds: 0.120, TFLOP/s/device: 113.078, Tokens/s/device: 17044.367, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0422 19:02:07.476298 125098811889472 metric_logger.py:196] completed step: 15, seconds: 0.118, TFLOP/s/device: 115.139, Tokens/s/device: 17355.050, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0422 19:02:07.592543 125098811889472 metric_logger.py:196] completed step: 16, seconds: 53.634, TFLOP/s/device: 0.253, Tokens/s/device: 38.185, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0422 19:02:07.711335 125098811889472 metric_logger.py:196] completed step: 17, seconds: 0.006, TFLOP/s/device: 2095.163, Tokens/s/device: 315805.705, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0422 19:02:07.830208 125098811889472 metric_logger.py:196] completed step: 18, seconds: 0.117, TFLOP/s/device: 115.649, Tokens/s/device: 17431.864, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0422 19:02:07.947955 125098811889472 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0422 19:02:07.949036 125098811889472 checkpointing.py:776] Waited 0.0011017322540283203 seconds for step 19 to finish before starting checkpointing. I0422 19:02:07.949308 125098811889472 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:02:08.244700 125069068076608 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_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0 I0422 19:02:56.738948 125068908693056 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 49.274522s (commit=48.775831s, array_metadata_write=0.498691s) I0422 19:02:56.740094 125066842998336 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 237.728 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 53 seconds) (per-host) I0422 19:02:56.740214 125066842998336 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.274503s. I0422 19:02:57.230404 125066842998336 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 19:02:57.668304 125066842998336 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 19:02:57.668489 125066842998336 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.613150s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15/items I0422 19:02:57.669397 125066842998336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15/items I0422 19:02:57.928704 125066842998336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15 I0422 19:02:58.651033 125066842998336 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_16_zarr3_true/checkpoints/15`. I0422 19:02:58.651776 125066842998336 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 55.161421s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/15 I0422 19:02:58.651840 125066842998336 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.186132s. I0422 19:02:58.651991 125066255795776 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:02:58.652127 125066255795776 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:02:58.652176 125066255795776 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:02:58.652213 125066255795776 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 19:02:58.652365 125098811889472 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 19:02:58.652492 125098811889472 checkpoint_manager.py:1441] Waiting for previous save to complete took 50.703193 seconds. If this number is high, consider checkpointing less frequently. I0422 19:02:58.652785 125098811889472 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0422 19:02:58.653115 125098811889472 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19. I0422 19:02:58.833802 125098811889472 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:02:58.833900 125098811889472 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 19:02:58.837403 125066842998336 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19 I0422 19:02:59.571178 125069068076608 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19/items I0422 19:03:05.194993 125098811889472 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.361606s I0422 19:03:05.199159 125098811889472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.912 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0422 19:03:05.199220 125098811889472 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.453980s (batch_requests_ready=0.085244s, total_serialization_initiated=6.364724s, others=0.004012s) I0422 19:03:05.199291 125098811889472 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.454544s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=6.454533) I0422 19:03:05.201089 125068908693056 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:03:05.201174 125098811889472 async_checkpointer.py:561] Finished blocking save. Time taken: 6.548349s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19. I0422 19:03:05.201338 125098811889472 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:03:05.201632 125068856264256 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:03:05.201755 125098811889472 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884527.9492848, 'wait_for_prev_duration_secs': 50.70319318771362, 'time_between_consecutive_saves_sec': 4.46065616607666, 'checkpointer_blocking_start_time': 1776884578.6528091, 'checkpointer_blocking_duration_secs': 6.548469066619873, 'get_old_steps_start_time': 1776884585.201288, 'get_old_steps_duration_secs': 1.8596649169921875e-05, 'checkpoint_manager_blocking_start_time': 1776884527.9492536, 'checkpoint_manager_blocking_duration_secs': 57.252477169036865} I0422 19:03:05.201868 125098811889472 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0422 19:03:05.201899 125098811889472 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:03:05.986478 125067457463872 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_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0 I0422 19:03:44.097505 125068877235776 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.897213s (commit=38.410927s, array_metadata_write=0.486287s) I0422 19:03:44.098425 125068908693056 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 278.631 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0422 19:03:44.098479 125068908693056 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.897263s. I0422 19:03:44.621388 125068908693056 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 19:03:45.022980 125068908693056 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 19:03:45.023225 125068908693056 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.588748s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19/items I0422 19:03:45.024106 125068908693056 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19/items I0422 19:03:45.274538 125068908693056 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19 I0422 19:03:46.049730 125068908693056 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_16_zarr3_true/checkpoints/19`. I0422 19:03:46.050485 125068908693056 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.397665s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_true/checkpoints/19 I0422 19:03:46.050544 125068908693056 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.849328s. I0422 19:03:46.050696 125068856264256 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:03:46.050813 125068856264256 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:03:46.050864 125068856264256 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:03:46.050904 125068856264256 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 19:03:46.051063 125098811889472 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 19:03:46.051214 125098811889472 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 19:03:46.052088 125098811889472 metric_logger.py:196] completed step: 19, seconds: 0.120, TFLOP/s/device: 113.243, Tokens/s/device: 17069.227, 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