2026-04-23 02:19:11.481916: 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 00:00:1776910751.494983 2315379 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776910751.498826 2315379 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776910751.510598 2315379 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776910751.510615 2315379 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776910751.510617 2315379 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776910751.510619 2315379 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'} 2026-04-23 02:19:15.206226: 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) I0423 02:19:15.341705 125750169347904 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 02:19:57.491628 125750169347904 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 02:19:57.491890 125750169347904 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 02:19:57.491938 125750169347904 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 I0423 02:19:57.491969 125750169347904 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 02:19:57.491991 125750169347904 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 02:19:57.492175 125750169347904 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 02:19:58.149707 125750169347904 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 02:19:58.149978 125750169347904 checkpointing.py:677] Setting up checkpoint logger... I0423 02:19:58.150022 125750169347904 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 02:19:58.150073 125750169347904 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 02:19:58.150495 125750169347904 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 0x725e349b5c10>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 02:19:59.812171 125750169347904 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 02:19:59.812560 125750169347904 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 0x725c5e3554c0>}, handler_registry=None I0423 02:19:59.812784 125750169347904 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x725c5e3554c0>` 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`. I0423 02:19:59.812823 125750169347904 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725cd89c4a70>` 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`. I0423 02:19:59.812853 125750169347904 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 0x725c5e3554c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x725c5e3554c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725cd89c4a70>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x725cd89c4a70>}). I0423 02:19:59.813143 125750169347904 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 02:19:59.813195 125750169347904 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x725c5e364900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 02:20:00.259733 125720504305216 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776910680582968185, 'commit_timestamp_nsecs': 1776910728144260627, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0423 02:20:00.342918 125720504305216 checkpoint_manager.py:1847] Missing metrics for step 9 I0423 02:20:00.393864 125750169347904 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints I0423 02:20:00.394415 125750169347904 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_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x725c5e355820> I0423 02:20:00.394529 125750169347904 checkpointing.py:301] Checkpoint manager created! I0423 02:20:00.776324 125750169347904 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 02:20:00.776427 125750169347904 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:20:01.058721 125750169347904 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:20:01.058817 125750169347904 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:20:01.070988 125750169347904 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:20:01.071037 125750169347904 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:20:01.090070 125750169347904 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:20:01.090133 125750169347904 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:20:01.102355 125750169347904 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:20:01.102406 125750169347904 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:20:01.114653 125750169347904 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:20:01.114705 125750169347904 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:20:01.134118 125750169347904 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 02:20:01.134179 125750169347904 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:20:01.149590 125750169347904 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 02:20:01.149642 125750169347904 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 02:20:01.299461 125750169347904 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 02:20:01.299563 125750169347904 checkpointing.py:581] restoring from this run's directory step 9 I0423 02:20:01.457498 125750169347904 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/9. I0423 02:20:02.037921 2316898 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 02:20:02.037999 2316898 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 02:20:10.892529 125750169347904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.360 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0423 02:20:10.893020 125750169347904 checkpointer.py:318] Finished restoring checkpoint in 9.52 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/9. I0423 02:20:10.893390 125750169347904 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776910801.3744724, 'checkpointer_duration_secs': 9.518834352493286, 'checkpoint_manager_start_time': 1776910801.2997696, 'checkpoint_manager_duration_secs': 9.593546867370605} [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 I0423 02:20:11.271696 125750169347904 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:20:11.271819 125750169347904 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) I0423 02:20:11.271862 125750169347904 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) I0423 02:20:11.271907 125750169347904 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') I0423 02:20:11.271948 125750169347904 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) I0423 02:20:11.271979 125750169347904 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) I0423 02:20:11.272019 125750169347904 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) I0423 02:20:11.272077 125750169347904 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') I0423 02:20:11.272109 125750169347904 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) I0423 02:20:11.272138 125750169347904 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) I0423 02:20:11.272173 125750169347904 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 02:20:11.272208 125750169347904 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 02:20:11.673912 125750169347904 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:20:11.674003 125750169347904 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:20:11.685556 125750169347904 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:20:11.685608 125750169347904 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:20:12.496027 125750169347904 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. I0423 02:20:12.499617 125750169347904 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 02:20:12.499820 125750169347904 metric_logger.py:301] number parameters: 1.104 billion I0423 02:20:15.002989 125750169347904 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0423 02:20:15.120033 125750169347904 checkpointing.py:776] Waited 0.11702322959899902 seconds for step 10 to finish before starting checkpointing. I0423 02:20:15.120685 125750169347904 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:20:15.120876 125750169347904 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0423 02:20:15.121212 125750169347904 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10. I0423 02:20:15.207248 125750169347904 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 02:20:15.289591 125718321169984 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10 I0423 02:20:15.295497 125750169347904 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 02:20:15.295594 125750169347904 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 I0423 02:20:16.022136 125718310684224 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/items I0423 02:20:16.179603 125718279226944 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776910815941104349, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776910815941104349, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0423 02:20:18.445955 125750169347904 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.150879s I0423 02:20:18.446545 125750169347904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.811 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0423 02:20:18.446599 125750169347904 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.237900s (batch_requests_ready=0.083667s, total_serialization_initiated=3.153786s, others=0.000447s) I0423 02:20:18.446662 125750169347904 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.238871s (all_items=0.000021s, per_item={'items': '0.00002122'}, temp_paths=3.238850) I0423 02:20:18.452177 125718470067776 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:20:18.452303 125750169347904 async_checkpointer.py:561] Finished blocking save. Time taken: 3.331379s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10. I0423 02:20:18.452529 125750169347904 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:20:18.452735 125718300198464 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:20:18.452811 125750169347904 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776910815.1206722, 'wait_for_prev_duration_secs': 4.076957702636719e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776910815.1208985, 'checkpointer_blocking_duration_secs': 3.3315508365631104, 'get_old_steps_start_time': 1776910818.4524662, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776910815.1205876, 'checkpoint_manager_blocking_duration_secs': 3.332195520401001} I0423 02:20:18.452897 125750169347904 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0423 02:20:18.452934 125750169347904 max_utils.py:750] Memstats: After params initialized: I0423 02:20:18.452988 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 02:20:18.453013 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 02:20:18.453034 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 02:20:18.453068 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 02:20:18.453087 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 02:20:18.453105 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 02:20:18.453123 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 02:20:18.453140 125750169347904 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 02:20:18.702190 125750169347904 metric_logger.py:196] completed step: 10, seconds: 2.503, TFLOP/s/device: 5.429, Tokens/s/device: 818.361, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0423 02:20:18.818912 125750169347904 metric_logger.py:196] completed step: 11, seconds: 3.697, TFLOP/s/device: 3.675, Tokens/s/device: 553.932, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0423 02:20:18.938207 125750169347904 metric_logger.py:196] completed step: 12, seconds: 0.006, TFLOP/s/device: 2265.654, Tokens/s/device: 341504.085, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0423 02:20:19.057397 125750169347904 metric_logger.py:196] completed step: 13, seconds: 0.118, TFLOP/s/device: 115.616, Tokens/s/device: 17426.969, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0423 02:20:19.062564 125750169347904 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0423 02:20:19.156277 125718289712704 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0423 02:20:19.294960 125750169347904 checkpointing.py:776] Waited 0.2323596477508545 seconds for step 15 to finish before starting checkpointing. I0423 02:20:19.296324 125750169347904 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:20:47.547478 125718310684224 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.095975s (commit=28.609803s, array_metadata_write=0.486172s) I0423 02:20:47.548785 125718470067776 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 390.748 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 32 seconds) (per-host) I0423 02:20:47.548911 125718470067776 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.096549s. I0423 02:20:48.020663 125718470067776 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 02:20:48.223912 125718279226944 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776910815941104349, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0423 02:20:48.403095 125718470067776 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:20:48.403283 125718470067776 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557010s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/items I0423 02:20:48.404132 125718470067776 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10/items I0423 02:20:48.635558 125718470067776 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10 I0423 02:20:49.310157 125718470067776 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10`. I0423 02:20:49.310881 125718470067776 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 34.189970s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/10 I0423 02:20:49.310959 125718470067776 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 30.858599s. I0423 02:20:49.311107 125718300198464 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:20:49.311159 125718300198464 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:20:49.311217 125718300198464 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:20:49.311260 125718300198464 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 02:20:49.315302 125750169347904 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. W0423 02:20:49.315439 125750169347904 checkpoint_manager.py:1441] Waiting for previous save to complete took 30.019113 seconds. If this number is high, consider checkpointing less frequently. I0423 02:20:49.316458 125750169347904 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0423 02:20:49.316769 125750169347904 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15. I0423 02:20:49.488188 125750169347904 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 02:20:49.488291 125750169347904 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 I0423 02:20:49.557401 125718470067776 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15 I0423 02:20:50.258335 125718289712704 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15/items I0423 02:20:51.590682 125750169347904 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.102900s I0423 02:20:51.595066 125750169347904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.628 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 02:20:51.595143 125750169347904 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.192786s (batch_requests_ready=0.081656s, total_serialization_initiated=2.106866s, others=0.004265s) I0423 02:20:51.595232 125750169347904 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.193388s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=2.193372) I0423 02:20:51.596970 125720334435904 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:20:51.597086 125750169347904 async_checkpointer.py:561] Finished blocking save. Time taken: 2.280581s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15. I0423 02:20:51.597285 125750169347904 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:20:51.597482 125718300198464 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:20:51.597616 125750169347904 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776910819.2962983, 'wait_for_prev_duration_secs': 30.019112825393677, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776910849.316483, 'checkpointer_blocking_duration_secs': 2.2807159423828125, 'get_old_steps_start_time': 1776910851.5972183, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776910819.2962253, 'checkpoint_manager_blocking_duration_secs': 32.301363945007324} I0423 02:20:51.597740 125750169347904 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0423 02:20:51.598516 125750169347904 metric_logger.py:196] completed step: 14, seconds: 0.120, TFLOP/s/device: 113.168, Tokens/s/device: 17057.853, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0423 02:20:51.607988 125750169347904 metric_logger.py:196] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.566, Tokens/s/device: 17268.565, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0423 02:20:51.724355 125750169347904 metric_logger.py:196] completed step: 16, seconds: 32.543, TFLOP/s/device: 0.418, Tokens/s/device: 62.932, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0423 02:20:51.844113 125750169347904 metric_logger.py:196] completed step: 17, seconds: 0.007, TFLOP/s/device: 2016.493, Tokens/s/device: 303947.759, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0423 02:20:51.962988 125750169347904 metric_logger.py:196] completed step: 18, seconds: 0.118, TFLOP/s/device: 115.434, Tokens/s/device: 17399.431, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0423 02:20:52.080776 125750169347904 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0423 02:20:52.081887 125750169347904 checkpointing.py:776] Waited 0.0011396408081054688 seconds for step 19 to finish before starting checkpointing. I0423 02:20:52.082160 125750169347904 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:20:52.353275 125720514790976 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0423 02:21:31.244547 125718289712704 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.648664s (commit=39.207721s, array_metadata_write=0.440943s) I0423 02:21:31.245677 125720334435904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 302.003 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0423 02:21:31.245800 125720334435904 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.648666s. I0423 02:21:31.729377 125720334435904 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 02:21:32.147646 125720334435904 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:21:32.147854 125720334435904 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.579238s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15/items I0423 02:21:32.148717 125720334435904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15/items I0423 02:21:32.375867 125720334435904 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15 I0423 02:21:33.052932 125720334435904 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15`. I0423 02:21:33.053727 125720334435904 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.737229s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/15 I0423 02:21:33.053809 125720334435904 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.456677s. I0423 02:21:33.054031 125718300198464 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:21:33.054163 125718300198464 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:21:33.054222 125718300198464 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:21:33.054261 125718300198464 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0423 02:21:33.054394 125750169347904 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. W0423 02:21:33.054506 125750169347904 checkpoint_manager.py:1441] Waiting for previous save to complete took 40.972356 seconds. If this number is high, consider checkpointing less frequently. I0423 02:21:33.054785 125750169347904 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0423 02:21:33.055098 125750169347904 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19. I0423 02:21:33.222411 125750169347904 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 02:21:33.222506 125750169347904 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 I0423 02:21:33.224944 125720334435904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19 I0423 02:21:33.916345 125720514790976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19/items I0423 02:21:38.982816 125750169347904 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.760791s I0423 02:21:38.986743 125750169347904 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.112 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0423 02:21:38.986804 125750169347904 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.843391s (batch_requests_ready=0.075741s, total_serialization_initiated=5.763865s, others=0.003785s) I0423 02:21:38.986876 125750169347904 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.843964s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=5.843952) I0423 02:21:38.988605 125718300198464 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:21:38.988691 125750169347904 async_checkpointer.py:561] Finished blocking save. Time taken: 5.933863s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19. I0423 02:21:38.988853 125750169347904 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:21:38.989056 125720292492864 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:21:38.989172 125750169347904 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776910852.082136, 'wait_for_prev_duration_secs': 40.97235608100891, 'time_between_consecutive_saves_sec': 2.770859718322754, 'checkpointer_blocking_start_time': 1776910893.0548093, 'checkpointer_blocking_duration_secs': 5.933983087539673, 'get_old_steps_start_time': 1776910898.9888017, 'get_old_steps_duration_secs': 2.002716064453125e-05, 'checkpoint_manager_blocking_start_time': 1776910852.0821042, 'checkpoint_manager_blocking_duration_secs': 46.90704607963562} I0423 02:21:38.989284 125750169347904 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0423 02:21:38.989314 125750169347904 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:21:39.724399 125720302978624 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0423 02:22:21.199122 125673916073536 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.211549s (commit=41.752070s, array_metadata_write=0.459479s) I0423 02:22:21.200421 125718300198464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 262.954 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0423 02:22:21.200543 125718300198464 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.211808s. I0423 02:22:21.676648 125718300198464 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 02:22:22.023886 125718300198464 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:22:22.024122 125718300198464 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.523662s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19/items I0423 02:22:22.024863 125718300198464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19/items I0423 02:22:22.254970 125718300198464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19 I0423 02:22:22.962901 125718300198464 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19`. I0423 02:22:22.963678 125718300198464 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.908857s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_15_ocdbt_true/checkpoints/19 I0423 02:22:22.963754 125718300198464 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.975022s. I0423 02:22:22.963975 125720292492864 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:22:22.964115 125720292492864 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:22:22.964176 125720292492864 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:22:22.964219 125720292492864 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 02:22:22.964458 125750169347904 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. I0423 02:22:22.964622 125750169347904 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:22:22.965688 125750169347904 metric_logger.py:196] completed step: 19, seconds: 0.120, TFLOP/s/device: 113.499, Tokens/s/device: 17107.868, 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