2026-04-15 21:32:22.696997: 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 21:32:22.813235 138466482510976 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0415 21:33:03.526155 138466482510976 max_utils.py:800] System Information: Jax Version: 0.9.2
I0415 21:33:03.526261 138466482510976 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0415 21:33:03.526296 138466482510976 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 21:33:03.526323 138466482510976 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0415 21:33:03.526346 138466482510976 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0415 21:33:03.526424 138466482510976 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0415 21:33:03.618066 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:04.008031 138466482510976 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0415 21:33:04.008579 138466482510976 checkpointing.py:677] Setting up checkpoint logger...
I0415 21:33:04.008630 138466482510976 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0415 21:33:04.008721 138466482510976 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0415 21:33:04.009754 138466482510976 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 0x7deea6900980>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0415 21:33:05.684189 138466482510976 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0415 21:33:05.684441 138466482510976 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 0x7de829f3d130>}, handler_registry=None
I0415 21:33:05.684881 138466482510976 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7de829f3d130>` 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 21:33:05.684928 138466482510976 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7de829f3e1b0>` 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 21:33:05.684961 138466482510976 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 0x7de829f3d130>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7de829f3d130>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7de829f3e1b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7de829f3e1b0>}).
I0415 21:33:05.685305 138466482510976 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0415 21:33:05.685366 138466482510976 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7de829f31300> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0415 21:33:06.132882 138360362370624 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288678575425084, 'commit_timestamp_nsecs': 1776288716013173234, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9/_CHECKPOINT_METADATA
I0415 21:33:06.228910 138360362370624 checkpoint_manager.py:1847] Missing metrics for step 9
I0415 21:33:06.234857 138466482510976 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints
I0415 21:33:06.235342 138466482510976 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_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7de829f3c260>
I0415 21:33:06.235446 138466482510976 checkpointing.py:301] Checkpoint manager created!
W0415 21:33:06.253428 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:06.435457 138466482510976 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0415 21:33:06.435558 138466482510976 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:33:06.529999 138466482510976 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:33:06.530088 138466482510976 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:33:06.544301 138466482510976 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:33:06.544348 138466482510976 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:33:06.564749 138466482510976 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 21:33:06.564806 138466482510976 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:33:06.578961 138466482510976 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 21:33:06.579010 138466482510976 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:33:06.593064 138466482510976 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 21:33:06.593111 138466482510976 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:33:06.614615 138466482510976 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0415 21:33:06.614675 138466482510976 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:33:06.632749 138466482510976 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0415 21:33:06.632799 138466482510976 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0415 21:33:06.830369 138466482510976 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0415 21:33:06.830476 138466482510976 checkpointing.py:581] restoring from this run's directory step 9
I0415 21:33:06.999761 138466482510976 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9.
I0415 21:33:07.535315 3250747 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0415 21:33:07.535382 3250747 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0415 21:33:13.724807 138466482510976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.932 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host)
I0415 21:33:13.725109 138466482510976 checkpointer.py:318] Finished restoring checkpoint in 6.80 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/9.
I0415 21:33:13.725200 138466482510976 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776288786.9220612, 'checkpointer_duration_secs': 6.8031005859375, 'checkpoint_manager_start_time': 1776288786.8307333, 'checkpoint_manager_duration_secs': 6.8944313526153564}
[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 21:33:14.367754 138466482510976 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0415 21:33:14.367860 138466482510976 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 21:33:14.367901 138466482510976 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 21:33:14.367943 138466482510976 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 21:33:14.367981 138466482510976 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 21:33:14.368008 138466482510976 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 21:33:14.368062 138466482510976 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 21:33:14.368108 138466482510976 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 21:33:14.368141 138466482510976 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 21:33:14.368167 138466482510976 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 21:33:14.368201 138466482510976 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0415 21:33:14.368235 138466482510976 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0415 21:33:15.028067 138466482510976 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:33:15.028153 138466482510976 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0415 21:33:15.041813 138466482510976 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:33:15.041863 138466482510976 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0415 21:33:15.700002 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:15.763994 138466482510976 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 21:33:15.764784 138466482510976 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0415 21:33:15.764943 138466482510976 metric_logger.py:289] number parameters: 1.104 billion
W0415 21:33:16.565413 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:16.633594 138466482510976 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0415 21:33:16.750352 138466482510976 checkpointing.py:776] Waited 0.11674928665161133 seconds for step 10 to finish before starting checkpointing.
I0415 21:33:16.750735 138466482510976 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:33:16.750905 138466482510976 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10
I0415 21:33:16.751213 138466482510976 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10.
I0415 21:33:16.836068 138466482510976 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0415 21:33:16.929916 138358831449664 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10
I0415 21:33:16.930289 138466482510976 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:33:16.930440 138466482510976 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 21:33:17.143502 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.171325 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.175956 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.180159 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.184131 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.187896 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.191783 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:17.195586 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:17.654982 138358820963904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10/items
I0415 21:33:17.780470 138358789506624 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288797549904046, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776288797549904046, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA
I0415 21:33:24.648497 138466482510976 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.718845s
I0415 21:33:24.652854 138466482510976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.579 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host)
I0415 21:33:24.652920 138466482510976 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.815968s (batch_requests_ready=0.090036s, total_serialization_initiated=7.721677s, others=0.004255s)
I0415 21:33:24.652998 138466482510976 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.816487s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=7.816469)
I0415 21:33:24.654222 138358841935424 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:33:24.654377 138466482510976 async_checkpointer.py:561] Finished blocking save. Time taken: 7.903426s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10.
I0415 21:33:24.654572 138466482510976 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:33:24.654725 138358810478144 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:33:24.654796 138466482510976 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288796.750723, 'wait_for_prev_duration_secs': 3.814697265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288796.7509336, 'checkpointer_blocking_duration_secs': 7.903566122055054, 'get_old_steps_start_time': 1776288804.6545155, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776288796.7506385, 'checkpoint_manager_blocking_duration_secs': 7.904132127761841}
I0415 21:33:24.654911 138466482510976 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0415 21:33:24.654957 138466482510976 max_utils.py:750]
Memstats: After params initialized:
I0415 21:33:24.655000 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0415 21:33:24.655026 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0415 21:33:24.655058 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0415 21:33:24.655076 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0415 21:33:24.655093 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0415 21:33:24.655109 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0415 21:33:24.655125 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0415 21:33:24.655140 138466482510976 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0415 21:33:24.660363 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:24.665893 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:33:24.670192 3249171 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:33:24.856170 138466482510976 metric_logger.py:185] completed step: 10, seconds: 0.868, TFLOP/s/device: 15.650, Tokens/s/device: 2358.999, total_weights: 16384, loss: 3.602
I0415 21:33:24.972563 138466482510976 metric_logger.py:185] completed step: 11, seconds: 8.221, TFLOP/s/device: 1.653, Tokens/s/device: 249.119, total_weights: 16384, loss: 3.101
I0415 21:33:25.091073 138466482510976 metric_logger.py:185] completed step: 12, seconds: 0.005, TFLOP/s/device: 2601.403, Tokens/s/device: 392111.813, total_weights: 16384, loss: 2.684
I0415 21:33:25.209763 138466482510976 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.957, Tokens/s/device: 17478.280, total_weights: 16384, loss: 2.348
I0415 21:33:25.214516 138466482510976 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0415 21:33:25.446192 138466482510976 checkpointing.py:776] Waited 0.23164844512939453 seconds for step 15 to finish before starting checkpointing.
I0415 21:33:25.446562 138466482510976 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:33:25.462692 138358799992384 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_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_0
I0415 21:33:55.228490 138358820963904 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.574897s (commit=30.158491s, array_metadata_write=0.416406s)
I0415 21:33:55.229650 138358841935424 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 329.146 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host)
I0415 21:33:55.229763 138358841935424 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.575341s.
I0415 21:33:55.649685 138358841935424 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 21:33:55.855300 138358789506624 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288797549904046, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA
I0415 21:33:56.072417 138358841935424 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:33:56.073099 138358841935424 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.561497s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10/items
I0415 21:33:56.073733 138358841935424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10/items
I0415 21:33:56.310362 138358841935424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10
I0415 21:33:56.946399 138358841935424 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_08_checkpoint_async_true/checkpoints/10`.
I0415 21:33:56.947017 138358841935424 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.196071s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/10
I0415 21:33:56.947110 138358841935424 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.292691s.
I0415 21:33:56.947288 138358810478144 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:33:56.947397 138358810478144 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:33:56.947465 138358810478144 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:33:56.947515 138358810478144 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0415 21:33:56.947891 138466482510976 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 21:33:56.948930 138466482510976 checkpoint_manager.py:1441] Waiting for previous save to complete took 31.502368 seconds. If this number is high, consider checkpointing less frequently.
I0415 21:33:56.949746 138466482510976 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15
I0415 21:33:56.950029 138466482510976 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15.
I0415 21:33:57.121071 138466482510976 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:33:57.121160 138466482510976 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 21:33:57.125357 138358810478144 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15
I0415 21:33:57.811159 138358779020864 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15/items
I0415 21:33:59.496822 138466482510976 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.376171s
I0415 21:33:59.497397 138466482510976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.009 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host)
I0415 21:33:59.497447 138466482510976 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.463762s (batch_requests_ready=0.083707s, total_serialization_initiated=2.379580s, others=0.000475s)
I0415 21:33:59.497514 138466482510976 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.464248s (all_items=0.000015s, per_item={'items': '0.00001550'}, temp_paths=2.464233)
I0415 21:33:59.502712 138360320427584 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:33:59.502801 138466482510976 async_checkpointer.py:561] Finished blocking save. Time taken: 2.553003s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15.
I0415 21:33:59.503030 138466482510976 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:33:59.503228 138358841935424 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:33:59.503340 138466482510976 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288805.4465373, 'wait_for_prev_duration_secs': 31.502368211746216, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288836.9497702, 'checkpointer_blocking_duration_secs': 2.5531647205352783, 'get_old_steps_start_time': 1776288839.5029557, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776288805.4464734, 'checkpoint_manager_blocking_duration_secs': 34.05683898925781}
I0415 21:33:59.503478 138466482510976 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0415 21:33:59.503949 138466482510976 metric_logger.py:185] completed step: 14, seconds: 0.118, TFLOP/s/device: 114.883, Tokens/s/device: 17316.457, total_weights: 16384, loss: 2.083
I0415 21:33:59.511597 138466482510976 metric_logger.py:185] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.018, Tokens/s/device: 17185.966, total_weights: 16384, loss: 1.880
I0415 21:33:59.627357 138466482510976 metric_logger.py:185] completed step: 16, seconds: 34.295, TFLOP/s/device: 0.396, Tokens/s/device: 59.717, total_weights: 16384, loss: 1.726
I0415 21:33:59.745532 138466482510976 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2379.949, Tokens/s/device: 358731.827, total_weights: 16384, loss: 1.610
I0415 21:33:59.864178 138466482510976 metric_logger.py:185] completed step: 18, seconds: 0.117, TFLOP/s/device: 116.370, Tokens/s/device: 17540.554, total_weights: 16384, loss: 1.521
I0415 21:33:59.982173 138466482510976 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0415 21:33:59.983139 138466482510976 checkpointing.py:776] Waited 0.0009937286376953125 seconds for step 19 to finish before starting checkpointing.
I0415 21:33:59.983377 138466482510976 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:34:00.250683 138360330913344 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_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_0
I0415 21:34:32.662591 138358779020864 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.160611s (commit=32.709868s, array_metadata_write=0.450743s)
I0415 21:34:32.663748 138360320427584 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 354.667 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 35 seconds) (per-host)
I0415 21:34:32.663855 138360320427584 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.160996s.
I0415 21:34:33.059532 138360320427584 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 21:34:33.511900 138360320427584 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:34:33.512698 138360320427584 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.587378s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15/items
I0415 21:34:33.513398 138360320427584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15/items
I0415 21:34:33.760684 138360320427584 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15
I0415 21:34:34.449700 138360320427584 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_08_checkpoint_async_true/checkpoints/15`.
I0415 21:34:34.450345 138360320427584 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 37.500561s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/15
I0415 21:34:34.450428 138360320427584 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.947572s.
I0415 21:34:34.450605 138358841935424 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:34:34.450703 138358841935424 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:34:34.450768 138358841935424 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:34:34.450811 138358841935424 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0415 21:34:34.451286 138466482510976 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 21:34:34.451402 138466482510976 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.468035 seconds. If this number is high, consider checkpointing less frequently.
I0415 21:34:34.451621 138466482510976 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19
I0415 21:34:34.451911 138466482510976 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19.
I0415 21:34:34.625797 138466482510976 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:34:34.625947 138358841935424 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19
I0415 21:34:34.626026 138466482510976 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 21:34:35.337398 138358820963904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19/items
I0415 21:34:42.891271 138466482510976 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.266173s
I0415 21:34:42.895617 138466482510976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.478 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host)
I0415 21:34:42.895674 138466482510976 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.352217s (batch_requests_ready=0.078151s, total_serialization_initiated=8.269826s, others=0.004241s)
I0415 21:34:42.895746 138466482510976 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.352754s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=8.352738)
I0415 21:34:42.897101 138358841935424 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:34:42.897195 138466482510976 async_checkpointer.py:561] Finished blocking save. Time taken: 8.445534s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19.
I0415 21:34:42.897430 138466482510976 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:34:42.897558 138360320427584 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:34:42.897644 138466482510976 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288839.9833512, 'wait_for_prev_duration_secs': 34.468034982681274, 'time_between_consecutive_saves_sec': 3.035820960998535, 'checkpointer_blocking_start_time': 1776288874.4516435, 'checkpointer_blocking_duration_secs': 8.445716619491577, 'get_old_steps_start_time': 1776288882.897371, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776288839.9833202, 'checkpoint_manager_blocking_duration_secs': 42.9143009185791}
I0415 21:34:42.897731 138466482510976 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0415 21:34:42.897763 138466482510976 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:34:43.667815 138360330913344 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_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_0
I0415 21:35:14.916121 138360288970304 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.019781s (commit=31.584532s, array_metadata_write=0.435249s)
I0415 21:35:14.917234 138358841935424 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 312.995 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host)
I0415 21:35:14.917340 138358841935424 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.020103s.
I0415 21:35:15.353511 138358841935424 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 21:35:15.780910 138358841935424 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:35:15.781650 138358841935424 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.584895s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19/items
I0415 21:35:15.782320 138358841935424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19/items
I0415 21:35:16.021689 138358841935424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19
I0415 21:35:16.714115 138358841935424 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_08_checkpoint_async_true/checkpoints/19`.
I0415 21:35:16.714725 138358841935424 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.263069s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_08_checkpoint_async_true/checkpoints/19
I0415 21:35:16.714792 138358841935424 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.817556s.
I0415 21:35:16.714904 138360320427584 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:35:16.714953 138360320427584 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:35:16.715000 138360320427584 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:35:16.715036 138360320427584 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0415 21:35:16.715522 138466482510976 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 21:35:16.715669 138466482510976 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:35:16.716635 138466482510976 metric_logger.py:185] completed step: 19, seconds: 0.118, TFLOP/s/device: 114.919, Tokens/s/device: 17321.876, 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