2026-04-16 12:59:19.063581: 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 12:59:19.180911 138288421792896 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 13:00:11.980923 138288421792896 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 13:00:11.981035 138288421792896 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 13:00:11.981127 138288421792896 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 13:00:11.981153 138288421792896 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0416 13:00:11.981177 138288421792896 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 13:00:11.981301 138288421792896 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 13:00:12.076587 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:12.469763 138288421792896 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 13:00:12.470144 138288421792896 checkpointing.py:677] Setting up checkpoint logger...
I0416 13:00:12.470192 138288421792896 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 13:00:12.470270 138288421792896 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 13:00:12.470990 138288421792896 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 0x7dc532952cf0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 13:00:14.133647 138288421792896 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0416 13:00:14.133875 138288421792896 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 0x7dc02c9ab860>}, handler_registry=None
I0416 13:00:14.134239 138288421792896 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc02c9ab860>` 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 13:00:14.134284 138288421792896 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dbeb5b46360>` 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 13:00:14.134314 138288421792896 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 0x7dc02c9ab860>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc02c9ab860>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dbeb5b46360>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dbeb5b46360>}).
I0416 13:00:14.134621 138288421792896 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 13:00:14.134682 138288421792896 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7dbeb5b39260> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 13:00:14.522641 138190916683328 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776344217178442596, 'commit_timestamp_nsecs': 1776344241859846083, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 13:00:14.603794 138190916683328 checkpoint_manager.py:1847] Missing metrics for step 0
I0416 13:00:14.708910 138288421792896 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints
I0416 13:00:14.709424 138288421792896 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_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dc07f3f85f0>
I0416 13:00:14.709521 138288421792896 checkpointing.py:301] Checkpoint manager created!
W0416 13:00:14.727123 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:14.909211 138288421792896 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 13:00:14.909312 138288421792896 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:00:14.999459 138288421792896 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 13:00:14.999537 138288421792896 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:00:15.013626 138288421792896 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 13:00:15.013677 138288421792896 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:00:15.034135 138288421792896 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 13:00:15.034191 138288421792896 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:00:15.048295 138288421792896 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 13:00:15.048343 138288421792896 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:00:15.062432 138288421792896 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 13:00:15.062483 138288421792896 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:00:15.083879 138288421792896 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 13:00:15.083934 138288421792896 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:00:15.109999 138288421792896 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 13:00:15.110134 138288421792896 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 13:00:15.293543 138288421792896 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 13:00:15.293650 138288421792896 checkpointing.py:581] restoring from this run's directory step 9
I0416 13:00:15.505613 138288421792896 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/9.
I0416 13:00:16.056755 3845924 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 13:00:16.056821 3845924 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 13:00:22.229363 138288421792896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.931 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host)
I0416 13:00:22.229909 138288421792896 checkpointer.py:318] Finished restoring checkpoint in 6.85 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/9.
I0416 13:00:22.230108 138288421792896 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776344415.3800197, 'checkpointer_duration_secs': 6.849963426589966, 'checkpoint_manager_start_time': 1776344415.2939413, 'checkpoint_manager_duration_secs': 6.936046838760376}
[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 13:00:23.009464 138288421792896 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 13:00:23.009601 138288421792896 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 13:00:23.009640 138288421792896 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 13:00:23.009685 138288421792896 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 13:00:23.009728 138288421792896 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 13:00:23.009756 138288421792896 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 13:00:23.009796 138288421792896 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 13:00:23.009836 138288421792896 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 13:00:23.009867 138288421792896 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 13:00:23.009893 138288421792896 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 13:00:23.009929 138288421792896 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 13:00:23.009964 138288421792896 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 13:00:23.686462 138288421792896 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 13:00:23.686569 138288421792896 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 13:00:23.700806 138288421792896 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 13:00:23.700868 138288421792896 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 13:00:24.623903 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:24.692118 138288421792896 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 13:00:24.692755 138288421792896 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 13:00:24.692929 138288421792896 metric_logger.py:289] number parameters: 1.104 billion
W0416 13:00:25.546251 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:25.617659 138288421792896 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0416 13:00:25.733758 138288421792896 checkpointing.py:776] Waited 0.11607646942138672 seconds for step 10 to finish before starting checkpointing.
I0416 13:00:25.734375 138288421792896 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 13:00:25.734602 138288421792896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10
I0416 13:00:25.735062 138288421792896 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10.
I0416 13:00:25.829202 138288421792896 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 13:00:25.923524 138288421792896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 13:00:25.923646 138288421792896 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 13:00:25.931429 138189352207936 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10
W0416 13:00:26.163212 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.191580 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.196894 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.201471 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.205758 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.209922 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.214174 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:26.218478 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:26.659190 138189341722176 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10/items
I0416 13:00:26.830163 138189310264896 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776344426587892269, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776344426587892269, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA
I0416 13:00:35.661884 138288421792896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.738935s
I0416 13:00:35.667442 138288421792896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.255 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host)
I0416 13:00:35.667521 138288421792896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.836660s (batch_requests_ready=0.088860s, total_serialization_initiated=9.742479s, others=0.005322s)
I0416 13:00:35.668017 138288421792896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.838176s (all_items=0.000030s, per_item={'items': '0.00003004'}, temp_paths=9.838145)
I0416 13:00:35.669823 138189331236416 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 13:00:35.669934 138288421792896 async_checkpointer.py:561] Finished blocking save. Time taken: 9.935263s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10.
I0416 13:00:35.670244 138288421792896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 13:00:35.670493 138189362693696 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 13:00:35.670645 138288421792896 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776344425.7343588, 'wait_for_prev_duration_secs': 4.673004150390625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776344425.734625, 'checkpointer_blocking_duration_secs': 9.935454368591309, 'get_old_steps_start_time': 1776344435.6701238, 'get_old_steps_duration_secs': 4.9591064453125e-05, 'checkpoint_manager_blocking_start_time': 1776344425.734222, 'checkpoint_manager_blocking_duration_secs': 9.936383485794067}
I0416 13:00:35.670769 138288421792896 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0416 13:00:35.670839 138288421792896 max_utils.py:750]
Memstats: After params initialized:
I0416 13:00:35.670922 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0416 13:00:35.670952 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 13:00:35.670978 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 13:00:35.670998 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 13:00:35.671019 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 13:00:35.671039 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 13:00:35.671074 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 13:00:35.671093 138288421792896 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 13:00:35.678913 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:35.685851 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:00:35.690519 3844350 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:00:35.893422 138288421792896 metric_logger.py:185] completed step: 10, seconds: 0.924, TFLOP/s/device: 14.703, Tokens/s/device: 2216.122, total_weights: 16384, loss: 3.602
I0416 13:00:36.009341 138288421792896 metric_logger.py:185] completed step: 11, seconds: 10.274, TFLOP/s/device: 1.323, Tokens/s/device: 199.345, total_weights: 16384, loss: 3.101
I0416 13:00:36.128100 138288421792896 metric_logger.py:185] completed step: 12, seconds: 0.006, TFLOP/s/device: 2164.590, Tokens/s/device: 326270.511, total_weights: 16384, loss: 2.684
I0416 13:00:36.246456 138288421792896 metric_logger.py:185] completed step: 13, seconds: 0.117, TFLOP/s/device: 115.938, Tokens/s/device: 17475.446, total_weights: 16384, loss: 2.348
I0416 13:00:36.252009 138288421792896 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0416 13:00:36.385195 138189320750656 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_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_0
I0416 13:00:36.482704 138288421792896 checkpointing.py:776] Waited 0.23064732551574707 seconds for step 15 to finish before starting checkpointing.
I0416 13:00:36.483256 138288421792896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 13:01:12.403402 138189341722176 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.735187s (commit=36.245161s, array_metadata_write=0.490026s)
I0416 13:01:12.404853 138189331236416 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 271.327 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 46 seconds) (per-host)
I0416 13:01:12.404993 138189331236416 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.735119s.
I0416 13:01:12.856502 138189331236416 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 13:01:13.065966 138189310264896 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776344426587892269, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10/_CHECKPOINT_METADATA
I0416 13:01:13.306171 138189331236416 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 13:01:13.306421 138189331236416 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.593741s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10/items
I0416 13:01:13.307295 138189331236416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10/items
I0416 13:01:13.566719 138189331236416 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10
I0416 13:01:14.257860 138189331236416 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_08_checkpoint_async_true/checkpoints/10`.
I0416 13:01:14.258768 138189331236416 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.524128s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/10
I0416 13:01:14.258855 138189331236416 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.588984s.
I0416 13:01:14.259098 138189362693696 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 13:01:14.259246 138189362693696 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 13:01:14.259321 138189362693696 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0416 13:01:14.259374 138189362693696 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0416 13:01:14.260160 138288421792896 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 13:01:14.260874 138288421792896 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.777622 seconds. If this number is high, consider checkpointing less frequently.
I0416 13:01:14.261789 138288421792896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15
I0416 13:01:14.262163 138288421792896 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15.
I0416 13:01:14.451006 138189362693696 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15
I0416 13:01:14.458163 138288421792896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 13:01:14.458286 138288421792896 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 13:01:15.197939 138189352207936 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15/items
I0416 13:01:19.635860 138288421792896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.178275s
I0416 13:01:19.640882 138288421792896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.336 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host)
I0416 13:01:19.640955 138288421792896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.281864s (batch_requests_ready=0.094800s, total_serialization_initiated=5.182214s, others=0.004850s)
I0416 13:01:19.641036 138288421792896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.282560s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=5.282540)
I0416 13:01:19.643128 138189362693696 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 13:01:19.643237 138288421792896 async_checkpointer.py:561] Finished blocking save. Time taken: 5.381401s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15.
I0416 13:01:19.643470 138288421792896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 13:01:19.643735 138190822311488 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 13:01:19.643885 138288421792896 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776344436.4832227, 'wait_for_prev_duration_secs': 37.77762174606323, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776344474.2618132, 'checkpointer_blocking_duration_secs': 5.381558418273926, 'get_old_steps_start_time': 1776344479.6433914, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776344436.4831355, 'checkpoint_manager_blocking_duration_secs': 43.16071701049805}
I0416 13:01:19.644021 138288421792896 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0416 13:01:19.644602 138288421792896 metric_logger.py:185] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.257, Tokens/s/device: 17222.096, total_weights: 16384, loss: 2.083
I0416 13:01:19.653059 138288421792896 metric_logger.py:185] completed step: 15, seconds: 0.118, TFLOP/s/device: 114.973, Tokens/s/device: 17329.937, total_weights: 16384, loss: 1.880
I0416 13:01:19.768797 138288421792896 metric_logger.py:185] completed step: 16, seconds: 43.399, TFLOP/s/device: 0.313, Tokens/s/device: 47.190, total_weights: 16384, loss: 1.726
I0416 13:01:19.887294 138288421792896 metric_logger.py:185] completed step: 17, seconds: 0.006, TFLOP/s/device: 2101.644, Tokens/s/device: 316782.676, total_weights: 16384, loss: 1.610
I0416 13:01:20.005692 138288421792896 metric_logger.py:185] completed step: 18, seconds: 0.122, TFLOP/s/device: 111.477, Tokens/s/device: 16803.000, total_weights: 16384, loss: 1.521
I0416 13:01:20.123918 138288421792896 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0416 13:01:20.125247 138288421792896 checkpointing.py:776] Waited 0.001359701156616211 seconds for step 19 to finish before starting checkpointing.
I0416 13:01:20.125561 138288421792896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 13:01:20.418970 138190874740288 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_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_0
I0416 13:01:48.898684 138189352207936 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.257187s (commit=28.816473s, array_metadata_write=0.440715s)
I0416 13:01:48.900032 138189362693696 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 365.850 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host)
I0416 13:01:48.900099 138189362693696 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.256806s.
I0416 13:01:49.339406 138189362693696 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 13:01:49.756930 138189362693696 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 13:01:49.757978 138189362693696 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573303s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15/items
I0416 13:01:49.758862 138189362693696 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15/items
I0416 13:01:50.000209 138189362693696 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15
I0416 13:01:50.718750 138189362693696 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_08_checkpoint_async_true/checkpoints/15`.
I0416 13:01:50.719627 138189362693696 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.457798s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/15
I0416 13:01:50.719721 138189362693696 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.076429s.
I0416 13:01:50.719983 138190822311488 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 13:01:50.720141 138190822311488 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 13:01:50.720216 138190822311488 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0416 13:01:50.720268 138190822311488 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0416 13:01:50.720578 138288421792896 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 13:01:50.721070 138288421792896 checkpoint_manager.py:1441] Waiting for previous save to complete took 30.595509 seconds. If this number is high, consider checkpointing less frequently.
I0416 13:01:50.721343 138288421792896 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19
I0416 13:01:50.721724 138288421792896 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19.
I0416 13:01:50.897892 138288421792896 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 13:01:50.898019 138288421792896 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 13:01:50.955821 138190822311488 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19
I0416 13:01:51.668236 138190832797248 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19/items
I0416 13:01:58.769765 138288421792896 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.872434s
I0416 13:01:58.774733 138288421792896 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.549 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host)
I0416 13:01:58.774799 138288421792896 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.967213s (batch_requests_ready=0.086058s, total_serialization_initiated=7.876352s, others=0.004804s)
I0416 13:01:58.774876 138288421792896 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.967887s (all_items=0.000029s, per_item={'items': '0.00002909'}, temp_paths=7.967858)
I0416 13:01:58.776885 138190811825728 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 13:01:58.777199 138288421792896 async_checkpointer.py:561] Finished blocking save. Time taken: 8.055810s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19.
I0416 13:01:58.777419 138288421792896 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 13:01:58.777613 138189362693696 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 13:01:58.777738 138288421792896 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776344480.1255267, 'wait_for_prev_duration_secs': 30.59550905227661, 'time_between_consecutive_saves_sec': 5.866137504577637, 'checkpointer_blocking_start_time': 1776344510.7213666, 'checkpointer_blocking_duration_secs': 8.055968284606934, 'get_old_steps_start_time': 1776344518.7773514, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776344480.1254828, 'checkpoint_manager_blocking_duration_secs': 38.6522262096405}
I0416 13:01:58.777877 138288421792896 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0416 13:01:58.777912 138288421792896 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 13:01:59.549199 138190874740288 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_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_0
I0416 13:02:21.653418 138190832797248 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 22.877662s (commit=22.434950s, array_metadata_write=0.442713s)
I0416 13:02:21.654778 138190811825728 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 409.659 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 30 seconds) (per-host)
I0416 13:02:21.654935 138190811825728 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 22.877681s.
I0416 13:02:22.073504 138190811825728 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 13:02:22.455803 138190811825728 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 13:02:22.456844 138190811825728 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.531788s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19/items
I0416 13:02:22.457688 138190811825728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19/items
I0416 13:02:22.705252 138190811825728 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19
I0416 13:02:23.382907 138190811825728 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_08_checkpoint_async_true/checkpoints/19`.
I0416 13:02:23.383792 138190811825728 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 32.662413s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_08_checkpoint_async_true/checkpoints/19
I0416 13:02:23.383861 138190811825728 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 24.606610s.
I0416 13:02:23.384040 138189362693696 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 13:02:23.384186 138189362693696 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 13:02:23.384235 138189362693696 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0416 13:02:23.384288 138189362693696 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0416 13:02:23.384460 138288421792896 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 13:02:23.384646 138288421792896 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 13:02:23.385594 138288421792896 metric_logger.py:185] completed step: 19, seconds: 0.114, TFLOP/s/device: 119.225, Tokens/s/device: 17970.903, 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