2026-04-22 18:54:44.998453: 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 00:00:1776884085.012253 2088024 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776884085.016310 2088024 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776884085.028884 2088024 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884085.028901 2088024 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884085.028903 2088024 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884085.028904 2088024 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 18:54:48.698323: 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 18:54:48.833236 127614405007168 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 18:55:46.203424 127614405007168 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:55:46.203691 127614405007168 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:55:46.203741 127614405007168 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 18:55:46.203772 127614405007168 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 18:55:46.203797 127614405007168 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:55:46.203959 127614405007168 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 18:55:46.859940 127614405007168 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:55:46.860203 127614405007168 checkpointing.py:677] Setting up checkpoint logger... I0422 18:55:46.860248 127614405007168 checkpointing.py:233] Creating checkpoint manager with ocdbt=False and zarr3=True I0422 18:55:46.860284 127614405007168 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 18:55:46.860724 127614405007168 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x741041baae40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:55:48.536568 127614405007168 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 18:55:48.536934 127614405007168 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 0x740ee4578380>}, handler_registry=None I0422 18:55:48.537166 127614405007168 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x740ee4578380>` 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 18:55:48.537208 127614405007168 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x740e76bda600>` 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 18:55:48.537237 127614405007168 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 0x740ee4578380>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x740ee4578380>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x740e76bda600>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x740e76bda600>}). I0422 18:55:48.537499 127614405007168 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 18:55:48.537548 127614405007168 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x740e76bf8900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:55:49.056804 127584799032896 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884021923799415, 'commit_timestamp_nsecs': 1776884030851504609, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0422 18:55:49.137383 127584799032896 checkpoint_manager.py:1847] Missing metrics for step 0 I0422 18:55:49.163375 127614405007168 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints I0422 18:55:49.163830 127614405007168 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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x740e76bda450> I0422 18:55:49.163940 127614405007168 checkpointing.py:301] Checkpoint manager created! I0422 18:55:49.553099 127614405007168 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:55:49.553193 127614405007168 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:55:49.834010 127614405007168 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:55:49.834104 127614405007168 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:55:49.846317 127614405007168 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:55:49.846365 127614405007168 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:55:49.865363 127614405007168 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 18:55:49.865418 127614405007168 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:55:49.877609 127614405007168 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 18:55:49.877660 127614405007168 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:55:49.889815 127614405007168 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 18:55:49.889864 127614405007168 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:55:49.909117 127614405007168 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:55:49.909173 127614405007168 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:55:49.924395 127614405007168 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:55:49.924448 127614405007168 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 18:55:50.074124 127614405007168 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:55:50.074216 127614405007168 checkpointing.py:581] restoring from this run's directory step 9 I0422 18:55:50.241148 127614405007168 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/9. I0422 18:55:51.163395 2089645 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 18:55:51.163449 2089645 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 18:55:57.278519 127614405007168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.858 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0422 18:55:57.278767 127614405007168 checkpointer.py:318] Finished restoring checkpoint in 7.11 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/9. I0422 18:55:57.278832 127614405007168 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776884150.1637492, 'checkpointer_duration_secs': 7.115060806274414, 'checkpoint_manager_start_time': 1776884150.074421, 'checkpoint_manager_duration_secs': 7.204391002655029} [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 18:55:57.614662 127614405007168 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 18:55:57.614773 127614405007168 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 18:55:57.614815 127614405007168 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 18:55:57.614859 127614405007168 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 18:55:57.614907 127614405007168 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 18:55:57.614937 127614405007168 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 18:55:57.614979 127614405007168 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 18:55:57.615020 127614405007168 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 18:55:57.615066 127614405007168 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 18:55:57.615096 127614405007168 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 18:55:57.615133 127614405007168 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 18:55:57.615168 127614405007168 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 18:55:58.013247 127614405007168 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:55:58.013330 127614405007168 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:55:58.024784 127614405007168 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:55:58.024834 127614405007168 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:55:58.788681 127614405007168 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 18:55:58.789241 127614405007168 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 18:55:58.789399 127614405007168 metric_logger.py:301] number parameters: 1.104 billion I0422 18:56:00.597233 127614405007168 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0422 18:56:00.714833 127614405007168 checkpointing.py:776] Waited 0.11758565902709961 seconds for step 10 to finish before starting checkpointing. I0422 18:56:00.715372 127614405007168 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:56:00.715517 127614405007168 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0422 18:56:00.715789 127614405007168 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_false/checkpoints/10. I0422 18:56:00.800799 127614405007168 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 18:56:00.886315 127614405007168 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:56:00.886452 127582764795456 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10 I0422 18:56:00.886530 127614405007168 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 18:56:01.636415 127582754309696 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/items I0422 18:56:01.800038 127582357947968 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884161547394775, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776884161547394775, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0422 18:56:02.988501 127614405007168 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.102708s I0422 18:56:02.992578 127614405007168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.634 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 18:56:02.992643 127614405007168 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.190537s (batch_requests_ready=0.081024s, total_serialization_initiated=2.105546s, others=0.003967s) I0422 18:56:02.992720 127614405007168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.191461s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=2.191441) I0422 18:56:02.994376 127582775281216 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:56:02.994462 127614405007168 async_checkpointer.py:561] Finished blocking save. Time taken: 2.278905s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10. I0422 18:56:02.994648 127614405007168 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:56:02.994807 127582764795456 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:56:02.994900 127614405007168 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884160.7153587, 'wait_for_prev_duration_secs': 4.1484832763671875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884160.715538, 'checkpointer_blocking_duration_secs': 2.27903151512146, 'get_old_steps_start_time': 1776884162.9945836, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776884160.7152772, 'checkpoint_manager_blocking_duration_secs': 2.279595375061035} I0422 18:56:02.995009 127614405007168 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0422 18:56:02.995064 127614405007168 max_utils.py:750] Memstats: After params initialized: I0422 18:56:02.995111 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 18:56:02.995137 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 18:56:02.995157 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 18:56:02.995174 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 18:56:02.995192 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 18:56:02.995209 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 18:56:02.995226 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 18:56:02.995241 127614405007168 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 18:56:03.263371 127614405007168 metric_logger.py:196] completed step: 10, seconds: 1.807, TFLOP/s/device: 7.518, Tokens/s/device: 1133.165, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0422 18:56:03.378403 127614405007168 metric_logger.py:196] completed step: 11, seconds: 2.662, TFLOP/s/device: 5.103, Tokens/s/device: 769.207, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0422 18:56:03.498134 127614405007168 metric_logger.py:196] completed step: 12, seconds: 0.011, TFLOP/s/device: 1255.394, Tokens/s/device: 189226.647, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0422 18:56:03.627579 127614405007168 metric_logger.py:196] completed step: 13, seconds: 0.115, TFLOP/s/device: 118.450, Tokens/s/device: 17854.029, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0422 18:56:03.642363 127614405007168 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0422 18:56:03.743062 127582404085312 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_false/checkpoints/10/items/array_metadatas/process_0 I0422 18:56:03.856931 127614405007168 checkpointing.py:776] Waited 0.2145388126373291 seconds for step 15 to finish before starting checkpointing. I0422 18:56:03.858230 127614405007168 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:56:13.261411 127582754309696 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.267829s (commit=9.814360s, array_metadata_write=0.453469s) I0422 18:56:13.262258 127582775281216 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1014.178 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0422 18:56:13.262310 127582775281216 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.267805s. W0422 18:56:13.630080 127582775281216 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/items, I0422 18:56:13.750011 127582775281216 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 18:56:13.750134 127582775281216 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.186139s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/items I0422 18:56:13.750769 127582775281216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/items I0422 18:56:13.961452 127582357947968 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884161547394775, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0422 18:56:13.992115 127582775281216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10 I0422 18:56:14.703650 127582775281216 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_false/checkpoints/10`. I0422 18:56:14.704278 127582775281216 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 13.988729s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/10 I0422 18:56:14.704348 127582775281216 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 11.709844s. I0422 18:56:14.704536 127582764795456 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:56:14.704644 127582764795456 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:56:14.704696 127582764795456 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:56:14.704738 127582764795456 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 18:56:14.704890 127614405007168 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 18:56:14.705001 127614405007168 checkpoint_manager.py:1441] Waiting for previous save to complete took 10.846786 seconds. If this number is high, consider checkpointing less frequently. I0422 18:56:14.705924 127614405007168 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0422 18:56:14.706220 127614405007168 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_false/checkpoints/15. I0422 18:56:14.873829 127582775281216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15 I0422 18:56:14.883793 127614405007168 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:56:14.883886 127614405007168 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 18:56:15.567502 127582404085312 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15/items I0422 18:56:19.733200 127614405007168 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.849787s I0422 18:56:19.737104 127614405007168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.495 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0422 18:56:19.737162 127614405007168 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.946698s (batch_requests_ready=0.090180s, total_serialization_initiated=4.852733s, others=0.003785s) I0422 18:56:19.737241 127614405007168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.947441s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=4.947430) I0422 18:56:19.738903 127582764795456 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:56:19.738997 127614405007168 async_checkpointer.py:561] Finished blocking save. Time taken: 5.033030s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15. I0422 18:56:19.739207 127614405007168 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:56:19.739372 127582775281216 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:56:19.739476 127614405007168 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884163.858199, 'wait_for_prev_duration_secs': 10.846786260604858, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884174.7059479, 'checkpointer_blocking_duration_secs': 5.033170461654663, 'get_old_steps_start_time': 1776884179.739137, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776884163.8581278, 'checkpoint_manager_blocking_duration_secs': 15.881324291229248} I0422 18:56:19.739633 127614405007168 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0422 18:56:19.740319 127614405007168 metric_logger.py:196] completed step: 14, seconds: 0.124, TFLOP/s/device: 109.314, Tokens/s/device: 16477.062, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0422 18:56:19.749168 127614405007168 metric_logger.py:196] completed step: 15, seconds: 0.133, TFLOP/s/device: 102.544, Tokens/s/device: 15456.604, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0422 18:56:19.865459 127614405007168 metric_logger.py:196] completed step: 16, seconds: 16.105, TFLOP/s/device: 0.844, Tokens/s/device: 127.167, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0422 18:56:19.986610 127614405007168 metric_logger.py:196] completed step: 17, seconds: 0.007, TFLOP/s/device: 2070.262, Tokens/s/device: 312052.415, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0422 18:56:20.104504 127614405007168 metric_logger.py:196] completed step: 18, seconds: 0.118, TFLOP/s/device: 115.376, Tokens/s/device: 17390.714, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0422 18:56:20.229397 127614405007168 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0422 18:56:20.235740 127614405007168 checkpointing.py:776] Waited 0.006372928619384766 seconds for step 19 to finish before starting checkpointing. I0422 18:56:20.236213 127614405007168 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:56:20.469157 127582550885952 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_false/checkpoints/15/items/array_metadatas/process_0 I0422 18:56:28.541248 127582404085312 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 8.803256s (commit=8.363374s, array_metadata_write=0.439882s) I0422 18:56:28.542271 127582764795456 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 918.923 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 13 seconds) (per-host) I0422 18:56:28.542378 127582764795456 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 8.803336s. W0422 18:56:28.867569 127582764795456 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15/items, I0422 18:56:28.993544 127582764795456 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 18:56:28.993655 127582764795456 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.175842s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15/items I0422 18:56:28.994268 127582764795456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15/items I0422 18:56:29.228111 127582764795456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15 I0422 18:56:29.901200 127582764795456 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_false/checkpoints/15`. I0422 18:56:29.901813 127582764795456 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 15.195853s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/15 I0422 18:56:29.901886 127582764795456 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 10.162845s. I0422 18:56:29.902094 127582775281216 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:56:29.902197 127582775281216 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:56:29.902262 127582775281216 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:56:29.902305 127582775281216 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0422 18:56:29.902472 127614405007168 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 18:56:29.902529 127614405007168 checkpoint_manager.py:1441] Waiting for previous save to complete took 9.666341 seconds. If this number is high, consider checkpointing less frequently. I0422 18:56:29.902864 127614405007168 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0422 18:56:29.903157 127614405007168 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_false/checkpoints/19. I0422 18:56:30.067259 127614405007168 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:56:30.067397 127582764795456 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19 I0422 18:56:30.067475 127614405007168 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 18:56:30.770970 127582550885952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19/items I0422 18:56:34.085704 127614405007168 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.018945s I0422 18:56:34.090008 127614405007168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.006 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0422 18:56:34.090262 127614405007168 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.105556s (batch_requests_ready=0.078465s, total_serialization_initiated=4.022697s, others=0.004394s) I0422 18:56:34.090368 127614405007168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.106316s (all_items=0.000010s, per_item={'items': '0.00000978'}, temp_paths=4.106306) I0422 18:56:34.091468 127582764795456 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:56:34.091549 127614405007168 async_checkpointer.py:561] Finished blocking save. Time taken: 4.188646s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19. I0422 18:56:34.091729 127614405007168 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:56:34.091954 127582775281216 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:56:34.092075 127614405007168 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884180.2361736, 'wait_for_prev_duration_secs': 9.666341304779053, 'time_between_consecutive_saves_sec': 5.531419992446899, 'checkpointer_blocking_start_time': 1776884189.902887, 'checkpointer_blocking_duration_secs': 4.188771724700928, 'get_old_steps_start_time': 1776884194.0916712, 'get_old_steps_duration_secs': 2.2649765014648438e-05, 'checkpoint_manager_blocking_start_time': 1776884180.2361255, 'checkpoint_manager_blocking_duration_secs': 13.855912923812866} I0422 18:56:34.092201 127614405007168 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0422 18:56:34.092232 127614405007168 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:56:34.946660 127582754309696 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_false/checkpoints/19/items/array_metadatas/process_0 I0422 18:56:41.128343 127582550885952 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 7.037668s (commit=6.622687s, array_metadata_write=0.414981s) I0422 18:56:41.129300 127582764795456 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1.107 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0422 18:56:41.129404 127582764795456 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 7.037812s. W0422 18:56:41.451257 127582764795456 ocdbt_utils.py:175] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19/items, I0422 18:56:41.569443 127582764795456 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 18:56:41.569552 127582764795456 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.173530s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19/items I0422 18:56:41.570183 127582764795456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19/items I0422 18:56:41.793511 127582764795456 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19 I0422 18:56:42.451442 127582764795456 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_false/checkpoints/19`. I0422 18:56:42.452070 127582764795456 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 12.549172s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_false/checkpoints/19 I0422 18:56:42.452131 127582764795456 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 8.360539s. I0422 18:56:42.452299 127582775281216 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:56:42.452403 127582775281216 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:56:42.452444 127582775281216 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:56:42.452480 127582775281216 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 18:56:42.452618 127614405007168 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 18:56:42.452756 127614405007168 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:56:42.453502 127614405007168 metric_logger.py:196] completed step: 19, seconds: 0.123, TFLOP/s/device: 110.410, Tokens/s/device: 16642.153, 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