2026-04-15 22:09:46.066366: 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) I0415 22:09:46.184098 132874307128448 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 22:10:39.534078 132874307128448 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 22:10:39.534187 132874307128448 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 22:10:39.534227 132874307128448 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0415 22:10:39.534260 132874307128448 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 22:10:39.534286 132874307128448 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 22:10:39.534391 132874307128448 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 22:10:39.627630 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:10:40.024876 132874307128448 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 22:10:40.025258 132874307128448 checkpointing.py:677] Setting up checkpoint logger... I0415 22:10:40.025306 132874307128448 checkpointing.py:233] Creating checkpoint manager with ocdbt=False and zarr3=True I0415 22:10:40.025380 132874307128448 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 22:10:40.026117 132874307128448 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 0x78d89ef04740>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 22:10:41.717219 132874307128448 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 22:10:41.717431 132874307128448 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 0x78d22313d490>}, handler_registry=None I0415 22:10:41.717814 132874307128448 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78d22313d490>` 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`. I0415 22:10:41.717860 132874307128448 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78d22313e240>` 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`. I0415 22:10:41.717890 132874307128448 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 0x78d22313d490>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78d22313d490>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78d22313e240>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78d22313e240>}). I0415 22:10:41.718204 132874307128448 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 22:10:41.718262 132874307128448 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x78d2231313a0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 22:10:42.176814 132768216385088 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776290912950238189, 'commit_timestamp_nsecs': 1776290919957480704, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0415 22:10:42.276058 132768216385088 checkpoint_manager.py:1847] Missing metrics for step 0 I0415 22:10:42.353255 132874307128448 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints I0415 22:10:42.353755 132874307128448 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_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78d22313e0f0> I0415 22:10:42.353862 132874307128448 checkpointing.py:301] Checkpoint manager created! W0415 22:10:42.375220 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:10:42.561145 132874307128448 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 22:10:42.561336 132874307128448 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:10:42.656788 132874307128448 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 22:10:42.656873 132874307128448 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:10:42.671178 132874307128448 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 22:10:42.671226 132874307128448 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:10:42.692357 132874307128448 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:10:42.692411 132874307128448 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:10:42.706898 132874307128448 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:10:42.706949 132874307128448 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:10:42.721322 132874307128448 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:10:42.721372 132874307128448 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:10:42.743388 132874307128448 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 22:10:42.743443 132874307128448 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:10:42.762124 132874307128448 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 22:10:42.762174 132874307128448 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 22:10:42.947326 132874307128448 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 22:10:42.947429 132874307128448 checkpointing.py:581] restoring from this run's directory step 9 I0415 22:10:43.107040 132874307128448 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/9. I0415 22:10:43.993755 3288225 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 22:10:43.993805 3288225 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 22:10:50.650443 132874307128448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.734 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0415 22:10:50.650862 132874307128448 checkpointer.py:318] Finished restoring checkpoint in 7.62 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/9. I0415 22:10:50.650964 132874307128448 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776291043.0312693, 'checkpointer_duration_secs': 7.6196489334106445, 'checkpoint_manager_start_time': 1776291042.9477203, 'checkpoint_manager_duration_secs': 7.703200578689575} [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 I0415 22:10:51.016268 132874307128448 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 22:10:51.016373 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 22:10:51.016414 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 22:10:51.016461 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 22:10:51.016501 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 22:10:51.016528 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 22:10:51.016565 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0415 22:10:51.016603 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0415 22:10:51.016632 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0415 22:10:51.016658 132874307128448 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0415 22:10:51.016692 132874307128448 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0415 22:10:51.016725 132874307128448 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0415 22:10:51.434063 132874307128448 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 22:10:51.434145 132874307128448 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 22:10:51.448553 132874307128448 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 22:10:51.448603 132874307128448 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 22:10:52.071625 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:10:52.146514 132874307128448 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. I0415 22:10:52.146997 132874307128448 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 22:10:52.147166 132874307128448 metric_logger.py:289] number parameters: 1.104 billion W0415 22:10:52.963805 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:10:53.035194 132874307128448 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0415 22:10:53.151693 132874307128448 checkpointing.py:776] Waited 0.1164863109588623 seconds for step 10 to finish before starting checkpointing. I0415 22:10:53.152116 132874307128448 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 22:10:53.152281 132874307128448 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0415 22:10:53.152557 132874307128448 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10. I0415 22:10:53.234116 132874307128448 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 22:10:53.310958 132766603675200 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10 I0415 22:10:53.318373 132874307128448 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:10:53.318459 132874307128448 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 W0415 22:10:53.542890 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.557960 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.571373 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.577040 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.581479 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.585751 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.589868 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:10:53.594034 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:10:54.024464 132766593189440 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/items I0415 22:10:54.190265 132766557537856 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776291053944906675, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776291053944906675, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0415 22:11:01.258409 132874307128448 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.940444s I0415 22:11:01.262603 132874307128448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.537 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0415 22:11:01.262669 132874307128448 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.027646s (batch_requests_ready=0.080254s, total_serialization_initiated=7.943310s, others=0.004082s) I0415 22:11:01.262739 132874307128448 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.028186s (all_items=0.000026s, per_item={'items': '0.00002599'}, temp_paths=8.028160) I0415 22:11:01.263946 132766603675200 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:11:01.264026 132874307128448 async_checkpointer.py:561] Finished blocking save. Time taken: 8.111706s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10. I0415 22:11:01.264213 132874307128448 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:11:01.264401 132766614160960 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:11:01.264502 132874307128448 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776291053.152104, 'wait_for_prev_duration_secs': 3.981590270996094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776291053.1523025, 'checkpointer_blocking_duration_secs': 8.11183762550354, 'get_old_steps_start_time': 1776291061.2641551, 'get_old_steps_duration_secs': 2.2411346435546875e-05, 'checkpoint_manager_blocking_start_time': 1776291053.1519904, 'checkpoint_manager_blocking_duration_secs': 8.11248779296875} I0415 22:11:01.264611 132874307128448 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0415 22:11:01.264653 132874307128448 max_utils.py:750] Memstats: After params initialized: I0415 22:11:01.264697 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0415 22:11:01.264724 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 22:11:01.264743 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 22:11:01.264760 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 22:11:01.264777 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 22:11:01.264794 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 22:11:01.264810 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 22:11:01.264826 132874307128448 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 22:11:01.270711 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:11:01.277001 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:11:01.281727 3286539 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:11:01.473277 132874307128448 metric_logger.py:185] completed step: 10, seconds: 0.888, TFLOP/s/device: 15.309, Tokens/s/device: 2307.465, total_weights: 16384, loss: 3.602 I0415 22:11:01.589509 132874307128448 metric_logger.py:185] completed step: 11, seconds: 8.436, TFLOP/s/device: 1.611, Tokens/s/device: 242.755, total_weights: 16384, loss: 3.101 I0415 22:11:01.708750 132874307128448 metric_logger.py:185] completed step: 12, seconds: 0.006, TFLOP/s/device: 2145.449, Tokens/s/device: 323385.441, total_weights: 16384, loss: 2.684 I0415 22:11:01.827512 132874307128448 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 116.199, Tokens/s/device: 17514.752, total_weights: 16384, loss: 2.348 I0415 22:11:01.837275 132874307128448 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0415 22:11:02.044646 132766572217920 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_0 I0415 22:11:02.067378 132874307128448 checkpointing.py:776] Waited 0.23007655143737793 seconds for step 15 to finish before starting checkpointing. I0415 22:11:02.067833 132874307128448 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:11:07.123342 132766593189440 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 5.860010s (commit=5.478130s, array_metadata_write=0.381880s) I0415 22:11:07.124293 132766603675200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 909.828 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 13 seconds) (per-host) I0415 22:11:07.124408 132766603675200 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 5.860323s. W0415 22:11:07.435285 132766603675200 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_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/items, I0415 22:11:07.547723 132766603675200 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:11:07.547827 132766603675200 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.170810s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/items I0415 22:11:07.548480 132766603675200 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/items I0415 22:11:07.716933 132766557537856 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776291053944906675, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0415 22:11:07.796412 132766603675200 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10 I0415 22:11:08.454270 132766603675200 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10`. I0415 22:11:08.454900 132766603675200 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 15.302584s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/10 I0415 22:11:08.454975 132766603675200 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 7.190892s. I0415 22:11:08.455134 132766614160960 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:11:08.455235 132766614160960 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:11:08.455298 132766614160960 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:11:08.455343 132766614160960 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0415 22:11:08.455716 132874307128448 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. W0415 22:11:08.455832 132874307128448 checkpoint_manager.py:1441] Waiting for previous save to complete took 6.388007 seconds. If this number is high, consider checkpointing less frequently. I0415 22:11:08.456519 132874307128448 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0415 22:11:08.456812 132874307128448 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15. I0415 22:11:08.621096 132766614160960 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15 I0415 22:11:08.635135 132874307128448 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:11:08.635225 132874307128448 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 I0415 22:11:09.351690 132766547052096 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15/items I0415 22:11:20.430571 132874307128448 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.795834s I0415 22:11:20.435095 132874307128448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.037 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0415 22:11:20.435487 132874307128448 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.896063s (batch_requests_ready=0.092509s, total_serialization_initiated=11.798809s, others=0.004745s) I0415 22:11:20.435566 132874307128448 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.896576s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=11.896562) I0415 22:11:20.436636 132766614160960 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:11:20.436718 132874307128448 async_checkpointer.py:561] Finished blocking save. Time taken: 11.980154s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15. I0415 22:11:20.436913 132874307128448 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:11:20.437125 132765739648576 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:11:20.437227 132874307128448 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776291062.0678072, 'wait_for_prev_duration_secs': 6.388007402420044, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776291068.4565437, 'checkpointer_blocking_duration_secs': 11.980280637741089, 'get_old_steps_start_time': 1776291080.4368382, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776291062.0677354, 'checkpoint_manager_blocking_duration_secs': 18.369466543197632} I0415 22:11:20.437494 132874307128448 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0415 22:11:20.437974 132874307128448 metric_logger.py:185] completed step: 14, seconds: 0.120, TFLOP/s/device: 113.560, Tokens/s/device: 17117.019, total_weights: 16384, loss: 2.083 I0415 22:11:20.446874 132874307128448 metric_logger.py:185] completed step: 15, seconds: 0.122, TFLOP/s/device: 110.987, Tokens/s/device: 16729.156, total_weights: 16384, loss: 1.880 I0415 22:11:20.562335 132874307128448 metric_logger.py:185] completed step: 16, seconds: 18.607, TFLOP/s/device: 0.730, Tokens/s/device: 110.064, total_weights: 16384, loss: 1.726 I0415 22:11:20.682723 132874307128448 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2157.372, Tokens/s/device: 325182.598, total_weights: 16384, loss: 1.610 I0415 22:11:20.801595 132874307128448 metric_logger.py:185] completed step: 18, seconds: 0.116, TFLOP/s/device: 116.713, Tokens/s/device: 17592.235, total_weights: 16384, loss: 1.521 I0415 22:11:20.919759 132874307128448 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0415 22:11:20.921027 132874307128448 checkpointing.py:776] Waited 0.0013668537139892578 seconds for step 19 to finish before starting checkpointing. I0415 22:11:20.921438 132874307128448 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:11:21.276972 132766582703680 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_0 I0415 22:11:27.016740 132766547052096 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.581315s (commit=5.988565s, array_metadata_write=0.592750s) I0415 22:11:27.017630 132766614160960 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 683.876 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 18 seconds) (per-host) I0415 22:11:27.017690 132766614160960 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.580924s. W0415 22:11:27.336851 132766614160960 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_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15/items, I0415 22:11:27.427119 132766614160960 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:11:27.427228 132766614160960 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.141890s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15/items I0415 22:11:27.427860 132766614160960 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15/items I0415 22:11:27.671472 132766614160960 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15 I0415 22:11:28.347098 132766614160960 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15`. I0415 22:11:28.347731 132766614160960 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 19.891172s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/15 I0415 22:11:28.347799 132766614160960 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 7.911037s. I0415 22:11:28.347951 132765739648576 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:11:28.348062 132765739648576 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:11:28.348116 132765739648576 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:11:28.348154 132765739648576 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0415 22:11:28.348283 132874307128448 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. W0415 22:11:28.348390 132874307128448 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.426970 seconds. If this number is high, consider checkpointing less frequently. I0415 22:11:28.348584 132874307128448 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0415 22:11:28.348866 132874307128448 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19. I0415 22:11:28.513781 132874307128448 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:11:28.513863 132874307128448 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 I0415 22:11:28.520495 132766614160960 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19 I0415 22:11:29.237997 132766547052096 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19/items I0415 22:11:32.710016 132874307128448 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.196572s I0415 22:11:32.710561 132874307128448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.885 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0415 22:11:32.710611 132874307128448 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.277199s (batch_requests_ready=0.076864s, total_serialization_initiated=4.199913s, others=0.000422s) I0415 22:11:32.710673 132874307128448 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.277689s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=4.277678) I0415 22:11:32.715947 132766572217920 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:11:32.716061 132874307128448 async_checkpointer.py:561] Finished blocking save. Time taken: 4.367424s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19. I0415 22:11:32.716299 132874307128448 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:11:32.716498 132765739648576 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:11:32.716604 132874307128448 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776291080.921406, 'wait_for_prev_duration_secs': 7.426969528198242, 'time_between_consecutive_saves_sec': 12.466047763824463, 'checkpointer_blocking_start_time': 1776291088.3486118, 'checkpointer_blocking_duration_secs': 4.367626905441284, 'get_old_steps_start_time': 1776291092.7162492, 'get_old_steps_duration_secs': 1.811981201171875e-05, 'checkpoint_manager_blocking_start_time': 1776291080.9213607, 'checkpoint_manager_blocking_duration_secs': 11.79522156715393} I0415 22:11:32.716793 132874307128448 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0415 22:11:32.716824 132874307128448 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:11:33.533014 132766582703680 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_0 I0415 22:11:38.979630 132766547052096 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 6.264320s (commit=5.825423s, array_metadata_write=0.438896s) I0415 22:11:38.980763 132766572217920 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 1.170 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0415 22:11:38.980877 132766572217920 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 6.264771s. W0415 22:11:39.291433 132766572217920 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_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19/items, I0415 22:11:39.381806 132766572217920 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:11:39.381911 132766572217920 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.141008s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19/items I0415 22:11:39.382562 132766572217920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19/items I0415 22:11:39.620461 132766572217920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19 I0415 22:11:40.314778 132766572217920 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19`. I0415 22:11:40.315417 132766572217920 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 11.966794s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_false/checkpoints/19 I0415 22:11:40.315485 132766572217920 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 7.599380s. I0415 22:11:40.315695 132765739648576 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:11:40.315805 132765739648576 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:11:40.315857 132765739648576 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:11:40.315900 132765739648576 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0415 22:11:40.316057 132874307128448 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. I0415 22:11:40.316201 132874307128448 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 22:11:40.317321 132874307128448 metric_logger.py:185] completed step: 19, seconds: 0.128, TFLOP/s/device: 106.368, Tokens/s/device: 16032.943, total_weights: 16384, loss: 1.449 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops