MaxView

← Back to run

Log Summary

XPK Start: Mon Apr 20 18:36:50 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-20 18:37:14.856224: 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)
I0420 18:37:15.034027 135545037080384 max_utils.py:273] Attempting to initialize the jax distributed system...
I0420 18:37:24.074018 135545037080384 distributed.py:149] Starting JAX distributed service on [::]:8482
I0420 18:37:24.076475 135545037080384 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-cciom-slice-job-0-0.mt-15-ocdbt-false-sv-cciom:8482
I0420 18:37:25.232435 135545037080384 max_utils.py:284] Jax distributed system initialized!
I0420 18:37:30.380842 135545037080384 max_utils.py:800] System Information: Jax Version: 0.9.2
I0420 18:37:30.380943 135545037080384 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0420 18:37:30.380984 135545037080384 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0420 18:37:30.381019 135545037080384 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0420 18:37:31.064706 135545037080384 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0420 18:37:31.175921 135545037080384 checkpointing.py:688] Setting up checkpoint logger...
I0420 18:37:31.176035 135545037080384 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True
I0420 18:37:31.176090 135545037080384 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0420 18:37:31.176573 135545037080384 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7b4668d14170>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0420 18:37:34.147592 135545037080384 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0420 18:37:34.147829 135545037080384 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b31d863e000>}, handler_registry=None
I0420 18:37:34.148068 135545037080384 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b31d863e000>` 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`.
I0420 18:37:34.148131 135545037080384 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b31f84a5760>` 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`.
I0420 18:37:34.148168 135545037080384 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 0x7b31d863e000>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b31d863e000>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b31f84a5760>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b31f84a5760>}).
I0420 18:37:34.148487 135545037080384 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0420 18:37:34.148558 135545037080384 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b31f839cfe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0420 18:37:35.691417 135545037080384 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints
I0420 18:37:36.575967 135545037080384 checkpoint_manager.py:929] [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_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, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b454022e630>
I0420 18:37:36.576154 135545037080384 checkpointing.py:302] Checkpoint manager created!
I0420 18:37:37.533920 135545037080384 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0420 18:37:37.534035 135545037080384 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0420 18:37:38.892482 135545037080384 maxtext_utils.py:1836]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0420 18:37:38.892596 135545037080384 maxtext_utils.py:1836]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0420 18:37:38.892655 135545037080384 maxtext_utils.py:1836]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0420 18:37:38.892698 135545037080384 maxtext_utils.py:1836]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0420 18:37:38.892734 135545037080384 maxtext_utils.py:1836]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0420 18:37:38.892767 135545037080384 maxtext_utils.py:1836]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0420 18:37:38.892802 135545037080384 maxtext_utils.py:1836]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0420 18:37:38.892836 135545037080384 maxtext_utils.py:1836]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0420 18:37:38.892869 135545037080384 maxtext_utils.py:1836]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0420 18:37:38.892900 135545037080384 maxtext_utils.py:1836]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0420 18:37:38.892932 135545037080384 maxtext_utils.py:1836]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)
I0420 18:37:38.892964 135545037080384 maxtext_utils.py:1836]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')

I0420 18:37:39.291109 135545037080384 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0420 18:37:39.291198 135545037080384 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 18:37:39.296502 135545037080384 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0420 18:37:39.296557 135545037080384 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 18:37:39.312718 135545037080384 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 18:37:39.312775 135545037080384 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 18:37:39.328302 135545037080384 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 18:37:39.328367 135545037080384 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 18:37:39.353679 135545037080384 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 18:37:39.353747 135545037080384 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 18:37:39.369375 135545037080384 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 18:37:39.369436 135545037080384 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 18:37:39.384916 135545037080384 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 18:37:39.384976 135545037080384 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 18:37:39.414732 135545037080384 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0420 18:37:39.414803 135545037080384 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 18:37:39.434375 135545037080384 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0420 18:37:39.434439 135545037080384 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0420 18:37:50.751720 135545037080384 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0420 18:37:50.761288 135545037080384 metric_logger.py:301] number parameters: 1.104 billion
I0420 18:38:02.364701 135545037080384 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0420 18:38:02.515353 135545037080384 checkpointing.py:798] Waited 0.1506350040435791 seconds for step 0 to finish before starting checkpointing.
I0420 18:38:02.517617 135545037080384 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 18:38:02.519410 135545037080384 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0
I0420 18:38:02.520910 135545037080384 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0.
I0420 18:38:03.943430 135545037080384 signaling_client.py:364] Using JaxDistributedSignalingClient
I0420 18:38:03.962993 135545037080384 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0420 18:38:03.963066 135545037080384 replica_slices.py:424] 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
I0420 18:38:04.234160 135545037080384 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272631s
I0420 18:38:04.235075 135545037080384 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.442 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28345704078674316 s) (per-host)
I0420 18:38:04.235161 135545037080384 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283561s (batch_requests_ready=0.002995s, total_serialization_initiated=0.279726s, others=0.000840s)
I0420 18:38:04.235246 135545037080384 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287613s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.287596)
I0420 18:38:04.236105 135545037080384 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.72 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0.
I0420 18:38:04.236442 135419775858432 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:58:04.236404
I0420 18:38:04.248997 135545037080384 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 18:38:04.249281 135418128578304 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0420 18:38:04.249445 135545037080384 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710282.517592, 'wait_for_prev_duration_secs': 6.651878356933594e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776710282.5194497, 'checkpointer_blocking_duration_secs': 1.7171471118927002, 'get_old_steps_start_time': 1776710284.2366197, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776710282.5157914, 'checkpoint_manager_blocking_duration_secs': 1.7336077690124512}
I0420 18:38:04.249571 135545037080384 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0420 18:38:04.249664 135545037080384 max_utils.py:750] 
Memstats: After params initialized:
I0420 18:38:04.249729 135545037080384 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0))
I0420 18:38:04.249772 135545037080384 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0))
I0420 18:38:04.249806 135545037080384 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0))
I0420 18:38:04.249840 135545037080384 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0))
I0420 18:38:04.614879 135545037080384 metric_logger.py:196] completed step: 0, seconds: 11.272, TFLOP/s/device: 1.205, Tokens/s/device: 181.696, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469
I0420 18:38:04.618122 135545037080384 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/tensorboard/'
I0420 18:38:04.693938 135419767465728 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0
I0420 18:38:05.040303 135545037080384 metric_logger.py:196] completed step: 1, seconds: 2.246, TFLOP/s/device: 6.048, Tokens/s/device: 911.642, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469
I0420 18:38:05.182145 135545037080384 metric_logger.py:196] completed step: 2, seconds: 0.426, TFLOP/s/device: 31.867, Tokens/s/device: 4803.272, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637
I0420 18:38:05.329798 135545037080384 metric_logger.py:196] completed step: 3, seconds: 0.012, TFLOP/s/device: 1087.057, Tokens/s/device: 163853.108, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932
I0420 18:38:05.341909 135545037080384 checkpointing.py:794] Waiting for step 5 to finish before checkpoint...
I0420 18:38:05.624447 135545037080384 checkpointing.py:798] Waited 0.28250598907470703 seconds for step 5 to finish before starting checkpointing.
I0420 18:38:05.627240 135545037080384 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 18:38:07.390423 135419740006144 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/items
I0420 18:38:07.540982 135417097615104 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710287118410464, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776710287118410464, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA
I0420 18:38:08.545440    2796 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0420 18:38:10.780693 135419192399616 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_0
I0420 18:38:19.772784 135418171356928 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 11.224292s (commit=9.671219s, array_metadata_write=1.553073s)
I0420 18:38:19.774230 135419775858432 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 99.832 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 15.822571992874146 s) (per-host)
I0420 18:38:19.774398 135419775858432 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 15.537835s.
I0420 18:38:20.427545 135419775858432 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710287118410464, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA
I0420 18:38:21.603555 135417097615104 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776710287118410464, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA
W0420 18:38:21.660292 135419775858432 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/items, 
I0420 18:38:22.515332 135419775858432 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000772s.
I0420 18:38:22.515616 135419775858432 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.533095s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/items
I0420 18:38:22.518099 135419775858432 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/items
I0420 18:38:23.048519 135419775858432 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0
I0420 18:38:24.440935 135419775858432 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0`.
I0420 18:38:24.442110 135419775858432 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 21.92 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0
I0420 18:38:24.443802 135419775858432 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 20.207235s.
I0420 18:38:24.444149 135418128578304 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 18:38:24.444317 135418128578304 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0420 18:38:24.444430 135418128578304 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0420 18:38:24.447119 135418128578304 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0420 18:38:24.447381 135545037080384 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0420 18:38:24.447548 135545037080384 checkpoint_manager.py:1452] Waiting for previous save to complete took 18.820323 seconds. If this number is high, consider checkpointing less frequently.
I0420 18:38:24.449398 135545037080384 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5
I0420 18:38:24.451862 135545037080384 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5.
I0420 18:38:24.757989 135545037080384 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0420 18:38:24.758149 135545037080384 replica_slices.py:424] 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
I0420 18:38:24.799405 135545037080384 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044358s
I0420 18:38:24.800415 135545037080384 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 24.223 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06368231773376465 s) (per-host)
I0420 18:38:24.800524 135545037080384 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.063817s (batch_requests_ready=0.006150s, total_serialization_initiated=0.056773s, others=0.000893s)
I0420 18:38:24.800657 135545037080384 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.068900s (all_items=0.000026s, per_item={'items': '0.00002575'}, temp_paths=0.068874)
I0420 18:38:24.801631 135545037080384 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5.
I0420 18:38:24.801975 135418128578304 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:58:24.801947
I0420 18:38:24.804344 135545037080384 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 18:38:24.804631 135418171356928 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0420 18:38:24.804783 135545037080384 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710285.627196, 'wait_for_prev_duration_secs': 18.8203227519989, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776710304.4494638, 'checkpointer_blocking_duration_secs': 0.3526630401611328, 'get_old_steps_start_time': 1776710304.8021553, 'get_old_steps_duration_secs': 4.076957702636719e-05, 'checkpoint_manager_blocking_start_time': 1776710285.6249459, 'checkpoint_manager_blocking_duration_secs': 19.179787635803223}
I0420 18:38:24.804998 135545037080384 checkpointing.py:409] Started an asynchronous checkpoint save for step 5
I0420 18:38:24.990720 135419775858432 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5
I0420 18:38:26.812611 135418699835136 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5/items
I0420 18:38:29.713869 135418685568768 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_0
I0420 18:38:38.650918 135418666264320 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.696538s (commit=9.508245s, array_metadata_write=1.188293s)
I0420 18:38:38.652110 135418128578304 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 113.515 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 13.915342330932617 s) (per-host)
I0420 18:38:38.652235 135418128578304 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 13.850152s.
W0420 18:38:40.237685 135418128578304 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5/items, 
I0420 18:38:40.677479 135418128578304 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000401s.
I0420 18:38:40.677710 135418128578304 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.160506s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5/items
I0420 18:38:40.680426 135418128578304 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5/items
I0420 18:38:41.203544 135418128578304 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5
I0420 18:38:42.568839 135418128578304 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5`.
I0420 18:38:42.569649 135418128578304 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.12 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/5
I0420 18:38:42.571136 135418128578304 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.769051s.
I0420 18:38:42.571279 135418171356928 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 18:38:42.571340 135418171356928 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0420 18:38:42.571403 135418171356928 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts...
I0420 18:38:42.574188 135418171356928 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts.
I0420 18:38:48.209770 135545037080384 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.757, Tokens/s/device: 14282.825, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436
I0420 18:38:48.222125 135545037080384 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.489, Tokens/s/device: 13940.982, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778
I0420 18:38:48.366340 135545037080384 metric_logger.py:196] completed step: 6, seconds: 42.879, TFLOP/s/device: 0.317, Tokens/s/device: 47.762, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984
I0420 18:38:48.513866 135545037080384 metric_logger.py:196] completed step: 7, seconds: 0.010, TFLOP/s/device: 1365.541, Tokens/s/device: 205829.146, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266
I0420 18:38:48.661383 135545037080384 metric_logger.py:196] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.167, Tokens/s/device: 14043.172, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479
I0420 18:38:48.809515 135545037080384 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0420 18:38:48.810252 135545037080384 checkpointing.py:798] Waited 0.0007491111755371094 seconds for step 9 to finish before starting checkpointing.
I0420 18:38:48.812290 135545037080384 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 18:38:48.813895 135545037080384 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9
I0420 18:38:48.815429 135545037080384 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9.
I0420 18:38:49.544625 135545037080384 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0420 18:38:49.544719 135545037080384 replica_slices.py:424] 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
I0420 18:38:49.572743 135545037080384 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029567s
I0420 18:38:49.573749 135545037080384 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.506 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03808283805847168 s) (per-host)
I0420 18:38:49.573829 135545037080384 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038183s (batch_requests_ready=0.002546s, total_serialization_initiated=0.034705s, others=0.000933s)
I0420 18:38:49.573924 135545037080384 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042290s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.042274)
I0420 18:38:49.574677 135545037080384 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9.
I0420 18:38:49.575028 135418171356928 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:58:49.574992
I0420 18:38:49.581119 135545037080384 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 18:38:49.581360 135418677176064 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0420 18:38:49.581520 135545037080384 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776710328.8122542, 'wait_for_prev_duration_secs': 7.271766662597656e-05, 'time_between_consecutive_saves_sec': 6.2380266189575195, 'checkpointer_blocking_start_time': 1776710328.813933, 'checkpointer_blocking_duration_secs': 0.7612535953521729, 'get_old_steps_start_time': 1776710329.5752068, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776710328.810497, 'checkpoint_manager_blocking_duration_secs': 0.7709908485412598}
I0420 18:38:49.581640 135545037080384 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0420 18:38:49.581687 135545037080384 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 18:38:49.800843 135418128578304 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9
I0420 18:38:51.575877 135419775858432 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9/items
I0420 18:38:54.507306 135418699847424 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_0
I0420 18:39:02.941760 135418685568768 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.234943s (commit=8.713362s, array_metadata_write=1.521580s)
I0420 18:39:02.942789 135418171356928 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 117.818 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 13.407126665115356 s) (per-host)
I0420 18:39:02.942854 135418171356928 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 13.367695s.
W0420 18:39:04.352838 135418171356928 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9/items, 
I0420 18:39:04.767526 135418171356928 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000291s.
I0420 18:39:04.767723 135418171356928 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.052523s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9/items
I0420 18:39:04.769035 135418171356928 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9/items
I0420 18:39:05.302214 135418171356928 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9
I0420 18:39:07.076142 135418171356928 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9`.
I0420 18:39:07.076950 135418171356928 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.26 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9
I0420 18:39:07.078312 135418171356928 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.503150s.
I0420 18:39:07.078527 135418677176064 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 18:39:07.078648 135418677176064 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0420 18:39:07.078701 135418677176064 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0420 18:39:07.080446 135418677176064 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0420 18:39:07.080586 135545037080384 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0420 18:39:07.080696 135545037080384 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 18:39:07.081637 135545037080384 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.352, Tokens/s/device: 13920.231, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Mon Apr 20 18:39:16 UTC 2026
EXIT_CODE=0