2026-04-16 13:45:44.263274: 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) I0416 13:45:44.379465 138329518083200 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 13:46:21.889828 138329518083200 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 13:46:21.889952 138329518083200 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 13:46:21.889986 138329518083200 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 I0416 13:46:21.890013 138329518083200 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0416 13:46:21.890038 138329518083200 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 13:46:21.890225 138329518083200 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 13:46:21.983818 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:22.390865 138329518083200 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 13:46:22.391209 138329518083200 checkpointing.py:677] Setting up checkpoint logger... I0416 13:46:22.391255 138329518083200 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 13:46:22.391295 138329518083200 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 13:46:22.391968 138329518083200 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 0x7dcec31f8ad0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 13:46:24.056984 138329518083200 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0416 13:46:24.057229 138329518083200 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 0x7dc846b39ca0>}, handler_registry=None I0416 13:46:24.057511 138329518083200 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc846b39ca0>` 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`. I0416 13:46:24.057554 138329518083200 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc846b3a180>` 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`. I0416 13:46:24.057596 138329518083200 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 0x7dc846b39ca0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc846b39ca0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc846b3a180>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc846b3a180>}). I0416 13:46:24.057894 138329518083200 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 13:46:24.057950 138329518083200 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7dc846b31300> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 13:46:24.473263 138223416247872 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776346997959337047, 'commit_timestamp_nsecs': 1776347025172562160, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0416 13:46:24.550669 138223416247872 checkpoint_manager.py:1847] Missing metrics for step 0 I0416 13:46:24.623904 138329518083200 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints I0416 13:46:24.624510 138329518083200 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_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dc846b3a0f0> I0416 13:46:24.624601 138329518083200 checkpointing.py:301] Checkpoint manager created! W0416 13:46:24.642129 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:24.835394 138329518083200 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 13:46:24.835561 138329518083200 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:46:24.925572 138329518083200 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 13:46:24.925659 138329518083200 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:46:24.939777 138329518083200 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 13:46:24.939826 138329518083200 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 13:46:24.960339 138329518083200 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 13:46:24.960396 138329518083200 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:46:24.974630 138329518083200 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 13:46:24.974684 138329518083200 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:46:24.988806 138329518083200 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 13:46:24.988857 138329518083200 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:46:25.026224 138329518083200 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 13:46:25.026347 138329518083200 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 13:46:25.045039 138329518083200 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 13:46:25.045110 138329518083200 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 13:46:25.230437 138329518083200 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0416 13:46:25.230542 138329518083200 checkpointing.py:581] restoring from this run's directory step 9 I0416 13:46:25.407385 138329518083200 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/9. I0416 13:46:25.925355 3889214 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 13:46:25.925457 3889214 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 13:46:32.109155 138329518083200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.935 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0416 13:46:32.109451 138329518083200 checkpointer.py:318] Finished restoring checkpoint in 6.80 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/9. I0416 13:46:32.109542 138329518083200 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776347185.312467, 'checkpointer_duration_secs': 6.797034502029419, 'checkpoint_manager_start_time': 1776347185.2307932, 'checkpoint_manager_duration_secs': 6.878711700439453} [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 I0416 13:46:32.595738 138329518083200 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 13:46:32.595859 138329518083200 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) I0416 13:46:32.595899 138329518083200 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) I0416 13:46:32.595943 138329518083200 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') I0416 13:46:32.595982 138329518083200 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) I0416 13:46:32.596010 138329518083200 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) I0416 13:46:32.596227 138329518083200 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) I0416 13:46:32.596279 138329518083200 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') I0416 13:46:32.596312 138329518083200 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) I0416 13:46:32.596338 138329518083200 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) I0416 13:46:32.596375 138329518083200 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 13:46:32.596410 138329518083200 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 13:46:33.017943 138329518083200 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 13:46:33.018037 138329518083200 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 13:46:33.031593 138329518083200 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 13:46:33.031643 138329518083200 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 13:46:33.665225 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:33.733390 138329518083200 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. I0416 13:46:33.733992 138329518083200 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 13:46:33.734155 138329518083200 metric_logger.py:289] number parameters: 1.104 billion W0416 13:46:34.572703 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:34.648458 138329518083200 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0416 13:46:34.764844 138329518083200 checkpointing.py:776] Waited 0.11636042594909668 seconds for step 10 to finish before starting checkpointing. I0416 13:46:34.765370 138329518083200 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 13:46:34.765552 138329518083200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0416 13:46:34.765888 138329518083200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10. I0416 13:46:34.844873 138329518083200 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 13:46:34.922449 138221893715520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10 I0416 13:46:34.932322 138329518083200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:46:34.932421 138329518083200 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 W0416 13:46:35.185286 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.213339 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.219317 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.224223 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.228490 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.232617 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.236725 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:35.240895 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:35.600493 138221883229760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/items I0416 13:46:35.751126 138221851772480 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776347195518485383, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776347195518485383, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0416 13:46:45.210739 138329518083200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.278932s I0416 13:46:45.211442 138329518083200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.191 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0416 13:46:45.211501 138329518083200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.365497s (batch_requests_ready=0.082898s, total_serialization_initiated=10.282037s, others=0.000561s) I0416 13:46:45.211579 138329518083200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.366118s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=10.366098) I0416 13:46:45.217271 138221904201280 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:46:45.217444 138329518083200 async_checkpointer.py:561] Finished blocking save. Time taken: 10.451851s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10. I0416 13:46:45.217722 138329518083200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:46:45.217953 138221841286720 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:46:45.218074 138329518083200 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347194.7653558, 'wait_for_prev_duration_secs': 4.267692565917969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776347194.765573, 'checkpointer_blocking_duration_secs': 10.452041387557983, 'get_old_steps_start_time': 1776347205.2176409, 'get_old_steps_duration_secs': 3.218650817871094e-05, 'checkpoint_manager_blocking_start_time': 1776347194.765255, 'checkpoint_manager_blocking_duration_secs': 10.452774286270142} I0416 13:46:45.218205 138329518083200 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0416 13:46:45.218246 138329518083200 max_utils.py:750] Memstats: After params initialized: I0416 13:46:45.218297 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 13:46:45.218321 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 13:46:45.218341 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 13:46:45.218359 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 13:46:45.218376 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 13:46:45.218393 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 13:46:45.218409 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 13:46:45.218425 138329518083200 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 13:46:45.225122 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:45.231979 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 13:46:45.236686 3887651 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 13:46:45.447825 138329518083200 metric_logger.py:185] completed step: 10, seconds: 0.914, TFLOP/s/device: 14.869, Tokens/s/device: 2241.159, total_weights: 16384, loss: 3.602 I0416 13:46:45.563904 138329518083200 metric_logger.py:185] completed step: 11, seconds: 10.798, TFLOP/s/device: 1.258, Tokens/s/device: 189.673, total_weights: 16384, loss: 3.101 I0416 13:46:45.682817 138329518083200 metric_logger.py:185] completed step: 12, seconds: 0.006, TFLOP/s/device: 2436.279, Tokens/s/device: 367222.521, total_weights: 16384, loss: 2.684 I0416 13:46:45.801301 138329518083200 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.894, Tokens/s/device: 17468.739, total_weights: 16384, loss: 2.348 I0416 13:46:45.806500 138329518083200 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0416 13:46:45.944032 138221862258240 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0 I0416 13:46:46.037839 138329518083200 checkpointing.py:776] Waited 0.23129606246948242 seconds for step 15 to finish before starting checkpointing. I0416 13:46:46.038318 138329518083200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:47:18.959994 138221883229760 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.743381s (commit=33.274662s, array_metadata_write=0.468719s) I0416 13:47:18.961468 138221904201280 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 286.448 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0416 13:47:18.961597 138221904201280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.744092s. I0416 13:47:19.405883 138221904201280 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 13:47:19.612088 138221851772480 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776347195518485383, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0416 13:47:19.834843 138221904201280 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 13:47:19.835626 138221904201280 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573023s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/items I0416 13:47:19.836385 138221904201280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10/items I0416 13:47:20.099836 138221904201280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10 I0416 13:47:20.760063 138221904201280 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10`. I0416 13:47:20.760805 138221904201280 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.995220s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/10 I0416 13:47:20.760890 138221904201280 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.543390s. I0416 13:47:20.761137 138221841286720 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:47:20.761266 138221841286720 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:47:20.761342 138221841286720 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:47:20.761394 138221841286720 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0416 13:47:20.761901 138329518083200 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. W0416 13:47:20.763385 138329518083200 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.725061 seconds. If this number is high, consider checkpointing less frequently. I0416 13:47:20.764359 138329518083200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0416 13:47:20.764675 138329518083200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15. I0416 13:47:20.935258 138329518083200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:47:20.935363 138329518083200 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 I0416 13:47:20.959956 138221841286720 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15 I0416 13:47:21.677537 138221872744000 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15/items I0416 13:47:26.191508 138329518083200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.256767s I0416 13:47:26.192139 138329518083200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.306 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0416 13:47:26.192196 138329518083200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.352114s (batch_requests_ready=0.090766s, total_serialization_initiated=5.260865s, others=0.000483s) I0416 13:47:26.192275 138329518083200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.352751s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=5.352735) I0416 13:47:26.198136 138221904201280 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:47:26.198226 138329518083200 async_checkpointer.py:561] Finished blocking save. Time taken: 5.433821s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15. I0416 13:47:26.198433 138329518083200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:47:26.198697 138223321876032 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:47:26.198834 138329518083200 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347206.0382924, 'wait_for_prev_duration_secs': 34.7250611782074, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776347240.7643828, 'checkpointer_blocking_duration_secs': 5.433955192565918, 'get_old_steps_start_time': 1776347246.198356, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776347206.0382226, 'checkpoint_manager_blocking_duration_secs': 40.160582065582275} I0416 13:47:26.198972 138329518083200 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0416 13:47:26.199650 138329518083200 metric_logger.py:185] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.142, Tokens/s/device: 17204.735, total_weights: 16384, loss: 2.083 I0416 13:47:26.208878 138329518083200 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 114.695, Tokens/s/device: 17288.098, total_weights: 16384, loss: 1.880 I0416 13:47:26.324340 138329518083200 metric_logger.py:185] completed step: 16, seconds: 40.400, TFLOP/s/device: 0.336, Tokens/s/device: 50.693, total_weights: 16384, loss: 1.726 I0416 13:47:26.442622 138329518083200 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2210.008, Tokens/s/device: 333116.461, total_weights: 16384, loss: 1.610 I0416 13:47:26.560940 138329518083200 metric_logger.py:185] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.576, Tokens/s/device: 17571.556, total_weights: 16384, loss: 1.521 I0416 13:47:26.678791 138329518083200 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0416 13:47:26.679765 138329518083200 checkpointing.py:776] Waited 0.0010080337524414062 seconds for step 19 to finish before starting checkpointing. I0416 13:47:26.680036 138329518083200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:47:26.971393 138223363819072 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0 I0416 13:47:59.468582 138221872744000 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.275761s (commit=32.801887s, array_metadata_write=0.473874s) I0416 13:47:59.469856 138221904201280 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 327.126 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0416 13:47:59.469982 138221904201280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.271699s. I0416 13:47:59.904609 138221904201280 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 13:48:00.358144 138221904201280 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 13:48:00.358900 138221904201280 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.588372s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15/items I0416 13:48:00.359745 138221904201280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15/items I0416 13:48:00.602479 138221904201280 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15 I0416 13:48:01.317144 138221904201280 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15`. I0416 13:48:01.317924 138221904201280 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.553516s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/15 I0416 13:48:01.317995 138221904201280 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.119720s. I0416 13:48:01.318245 138223321876032 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:48:01.318380 138223321876032 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:48:01.318459 138223321876032 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:48:01.318502 138223321876032 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0416 13:48:01.318884 138329518083200 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. W0416 13:48:01.319369 138329518083200 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.639336 seconds. If this number is high, consider checkpointing less frequently. I0416 13:48:01.319633 138329518083200 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0416 13:48:01.319968 138329518083200 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19. I0416 13:48:01.502170 138329518083200 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 13:48:01.502270 138329518083200 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 I0416 13:48:01.543327 138223321876032 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19 I0416 13:48:02.248851 138221883229760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19/items I0416 13:48:05.111601 138329518083200 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.609909s I0416 13:48:05.116263 138329518083200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.331 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0416 13:48:05.116327 138329518083200 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.705211s (batch_requests_ready=0.087228s, total_serialization_initiated=3.613459s, others=0.004524s) I0416 13:48:05.116402 138329518083200 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.705799s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=3.705782) I0416 13:48:05.118135 138223321876032 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 13:48:05.118226 138329518083200 async_checkpointer.py:561] Finished blocking save. Time taken: 3.798548s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19. I0416 13:48:05.118427 138329518083200 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0416 13:48:05.118674 138221904201280 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 13:48:05.118811 138329518083200 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776347246.6800122, 'wait_for_prev_duration_secs': 34.639336347579956, 'time_between_consecutive_saves_sec': 5.918602705001831, 'checkpointer_blocking_start_time': 1776347281.319658, 'checkpointer_blocking_duration_secs': 3.798670530319214, 'get_old_steps_start_time': 1776347285.1183507, 'get_old_steps_duration_secs': 3.361701965332031e-05, 'checkpoint_manager_blocking_start_time': 1776347246.6799777, 'checkpoint_manager_blocking_duration_secs': 38.438801527023315} I0416 13:48:05.118934 138329518083200 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0416 13:48:05.118970 138329518083200 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 13:48:05.861577 138223363819072 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0 I0416 13:48:34.759870 138221883229760 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.642741s (commit=29.198321s, array_metadata_write=0.444420s) I0416 13:48:34.760983 138223321876032 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 378.916 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0416 13:48:34.761040 138223321876032 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.642771s. I0416 13:48:35.156427 138223321876032 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 13:48:35.620123 138223321876032 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 13:48:35.620845 138223321876032 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.603564s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19/items I0416 13:48:35.621595 138223321876032 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19/items I0416 13:48:35.937084 138223321876032 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19 I0416 13:48:36.658501 138223321876032 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19`. I0416 13:48:36.659285 138223321876032 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.339613s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_16_zarr3_true/checkpoints/19 I0416 13:48:36.659361 138223321876032 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.541092s. I0416 13:48:36.659605 138221904201280 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 13:48:36.659735 138221904201280 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 13:48:36.659803 138221904201280 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0416 13:48:36.659854 138221904201280 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0416 13:48:36.660029 138329518083200 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. I0416 13:48:36.660199 138329518083200 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 13:48:36.660894 138329518083200 metric_logger.py:185] completed step: 19, seconds: 0.118, TFLOP/s/device: 114.867, Tokens/s/device: 17313.968, 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