XPK Start: Fri Apr 17 02:18:53 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 02:19:16.921587: 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) I0417 02:19:17.100251 140462700377920 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:19:26.140028 140462700377920 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:19:26.142306 140462700377920 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-rs-npju3-slice-job-0-0.mt-16-zarr3-true-rs-npju3:8482 I0417 02:19:27.249248 140462700377920 max_utils.py:284] Jax distributed system initialized! I0417 02:19:33.322722 140462700377920 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:19:33.322825 140462700377920 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:19:33.322893 140462700377920 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 I0417 02:19:33.322936 140462700377920 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:19:33.988720 140462700377920 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:19:34.101267 140462700377920 checkpointing.py:688] Setting up checkpoint logger... I0417 02:19:34.101379 140462700377920 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 02:19:34.101424 140462700377920 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:19:34.101617 140462700377920 base_pytree_checkpoint_handler.py:441] 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 0x7fbf6751c0e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:19:35.397575 140462700377920 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:19:35.397808 140462700377920 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 0x7fab107dedb0>}, handler_registry=None I0417 02:19:35.398061 140462700377920 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fab107dedb0>` 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`. I0417 02:19:35.398115 140462700377920 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7faaf85aa8d0>` 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`. I0417 02:19:35.398152 140462700377920 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 0x7fab107dedb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fab107dedb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7faaf85aa8d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7faaf85aa8d0>}). I0417 02:19:35.398476 140462700377920 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:19:35.398548 140462700377920 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7fab102ea020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:19:37.788473 140340782679808 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392203535045171, 'commit_timestamp_nsecs': 1776392248495303002, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0417 02:19:38.029676 140340782679808 checkpoint_manager.py:1871] Missing metrics for step 9 I0417 02:19:38.333763 140462700377920 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints I0417 02:19:38.901216 140462700377920 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7faae07ac950> I0417 02:19:38.901386 140462700377920 checkpointing.py:302] Checkpoint manager created! I0417 02:19:39.849363 140462700377920 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:19:39.849478 140462700377920 checkpointing.py:582] restoring from this run's directory step 9 I0417 02:19:40.097222 140462700377920 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9. I0417 02:19:40.774594 140462700377920 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9. I0417 02:19:40.775013 140462700377920 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.68 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9. I0417 02:19:42.597499 1459 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:19:48.592110 140462700377920 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 263.336 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.000998258590698 s) (per-host) I0417 02:19:48.609975 140462700377920 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 246.909 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.409659147262573 s) (per-host) I0417 02:19:50.437157 140462700377920 event_tracking.py:138] [process=0] [sync] Finished load in 10.34 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9 I0417 02:19:50.437310 140462700377920 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1776392380.097124, 'checkpointer_duration_secs': 10.340155363082886, 'checkpoint_manager_start_time': 1776392379.8501303, 'checkpoint_manager_duration_secs': 10.587151050567627} fsdp: 32 I0417 02:19:50.513891 140462700377920 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:19:50.514002 140462700377920 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:19:50.514052 140462700377920 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:19:50.514095 140462700377920 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:19:50.514137 140462700377920 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:19:50.514172 140462700377920 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:19:50.514209 140462700377920 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:19:50.514245 140462700377920 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:19:50.514280 140462700377920 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:19:50.514313 140462700377920 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:19:50.514347 140462700377920 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:19:50.514380 140462700377920 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:19:50.914936 140462700377920 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:19:50.915028 140462700377920 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:19:50.920486 140462700377920 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:19:50.920542 140462700377920 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:19:50.936717 140462700377920 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:19:50.936777 140462700377920 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:19:50.952183 140462700377920 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:19:50.952242 140462700377920 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:19:50.975815 140462700377920 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:19:50.975893 140462700377920 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:19:50.991421 140462700377920 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:19:50.991482 140462700377920 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:19:51.006938 140462700377920 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:19:51.006994 140462700377920 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:19:51.036363 140462700377920 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:19:51.036437 140462700377920 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:19:51.062502 140462700377920 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:19:51.062608 140462700377920 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:20:02.061408 140462700377920 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. I0417 02:20:02.070333 140462700377920 metric_logger.py:289] number parameters: 1.104 billion I0417 02:20:13.582214 140462700377920 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 02:20:13.733781 140462700377920 checkpointing.py:798] Waited 0.15155482292175293 seconds for step 10 to finish before starting checkpointing. I0417 02:20:13.736531 140462700377920 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:20:13.738558 140462700377920 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0417 02:20:13.740033 140462700377920 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10. I0417 02:20:14.452483 140462700377920 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:20:14.469722 140462700377920 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:20:14.469862 140462700377920 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 I0417 02:20:14.740213 140340178700032 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10 I0417 02:20:14.742080 140462700377920 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.273781s I0417 02:20:14.742931 140462700377920 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.463 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28238511085510254 s) (per-host) I0417 02:20:14.743009 140462700377920 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.282484s (batch_requests_ready=0.002582s, total_serialization_initiated=0.279131s, others=0.000771s) I0417 02:20:14.743097 140462700377920 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.286609s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.286592) I0417 02:20:14.743816 140462700377920 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10. I0417 02:20:14.744117 140340187092736 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:40:14.744083 I0417 02:20:14.754072 140462700377920 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:20:14.754352 140335883732736 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:20:14.754502 140462700377920 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392413.7365117, 'wait_for_prev_duration_secs': 6.318092346191406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392413.7385945, 'checkpointer_blocking_duration_secs': 1.0056700706481934, 'get_old_steps_start_time': 1776392414.7442877, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776392413.7341707, 'checkpoint_manager_blocking_duration_secs': 1.020296573638916} I0417 02:20:14.754603 140462700377920 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 02:20:14.754669 140462700377920 max_utils.py:750] Memstats: After params initialized: I0417 02:20:14.754716 140462700377920 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 02:20:14.754749 140462700377920 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 02:20:14.754776 140462700377920 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 02:20:14.754800 140462700377920 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 02:20:15.103341 140462700377920 metric_logger.py:185] completed step: 10, seconds: 11.197, TFLOP/s/device: 1.214, Tokens/s/device: 182.913, total_weights: 65536, loss: 6.827 I0417 02:20:15.244411 140462700377920 metric_logger.py:185] completed step: 11, seconds: 1.518, TFLOP/s/device: 8.951, Tokens/s/device: 1349.250, total_weights: 65536, loss: 6.522 I0417 02:20:15.392080 140462700377920 metric_logger.py:185] completed step: 12, seconds: 0.013, TFLOP/s/device: 1011.022, Tokens/s/device: 152392.291, total_weights: 65536, loss: 6.260 I0417 02:20:15.539829 140462700377920 metric_logger.py:185] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.430, Tokens/s/device: 14384.245, total_weights: 65536, loss: 6.039 I0417 02:20:15.553165 140462700377920 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 02:20:15.834944 140462700377920 checkpointing.py:798] Waited 0.281757116317749 seconds for step 15 to finish before starting checkpointing. I0417 02:20:15.837783 140462700377920 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:20:16.475157 140336392296192 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/items I0417 02:20:16.644596 140335875340032 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392416217919956, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776392416217919956, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0417 02:20:19.882850 140335900518144 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/items/array_metadatas/process_0 I0417 02:20:59.074229 140335892125440 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.435896s (commit=40.201739s, array_metadata_write=1.234157s) I0417 02:20:59.075767 140340187092736 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.405 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.61518335342407 s) (per-host) I0417 02:20:59.075961 140340187092736 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.331732s. I0417 02:21:01.007527 140335875340032 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392416217919956, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/_CHECKPOINT_METADATA I0417 02:21:01.112502 140340187092736 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000758s. I0417 02:21:01.609180 140340187092736 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:21:02.560859 140340187092736 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.204551s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/items I0417 02:21:02.562908 140340187092736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10/items I0417 02:21:03.130643 140340187092736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10 I0417 02:21:04.565185 140340187092736 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10`. I0417 02:21:04.566358 140340187092736 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.83 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/10 I0417 02:21:04.568006 140340187092736 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.823774s. I0417 02:21:04.568298 140335883732736 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:21:04.568456 140335883732736 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:21:04.568565 140335883732736 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:21:04.569976 140335883732736 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 02:21:04.570275 140462700377920 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0417 02:21:04.570453 140462700377920 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.732671 seconds. If this number is high, consider checkpointing less frequently. I0417 02:21:04.572107 140462700377920 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0417 02:21:04.574410 140462700377920 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15. I0417 02:21:05.309211 140462700377920 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:21:05.309347 140462700377920 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 I0417 02:21:05.341956 140462700377920 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035720s I0417 02:21:05.342796 140462700377920 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 28.157 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.054784297943115234 s) (per-host) I0417 02:21:05.342913 140462700377920 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.054922s (batch_requests_ready=0.006249s, total_serialization_initiated=0.047891s, others=0.000782s) I0417 02:21:05.343039 140462700377920 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.059777s (all_items=0.000045s, per_item={'items': '0.00004530'}, temp_paths=0.059732) I0417 02:21:05.344021 140462700377920 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15. I0417 02:21:05.344370 140335883732736 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:41:05.344333 I0417 02:21:05.346794 140462700377920 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:21:05.347116 140335900518144 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:21:05.347302 140462700377920 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392415.837739, 'wait_for_prev_duration_secs': 48.73267126083374, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392464.5721717, 'checkpointer_blocking_duration_secs': 0.7723755836486816, 'get_old_steps_start_time': 1776392465.3445773, 'get_old_steps_duration_secs': 4.220008850097656e-05, 'checkpoint_manager_blocking_start_time': 1776392415.835444, 'checkpoint_manager_blocking_duration_secs': 49.51180601119995} I0417 02:21:05.347453 140462700377920 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 02:21:05.552160 140340187092736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15 I0417 02:21:07.325937 140335908910848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15/items I0417 02:21:10.272195 140339092379392 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15/items/array_metadatas/process_0 I0417 02:21:27.010053 140462700377920 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.509, Tokens/s/device: 13944.020, total_weights: 65536, loss: 5.858 I0417 02:21:27.021027 140462700377920 metric_logger.py:185] completed step: 15, seconds: 0.149, TFLOP/s/device: 91.357, Tokens/s/device: 13770.382, total_weights: 65536, loss: 5.713 I0417 02:21:27.165140 140462700377920 metric_logger.py:185] completed step: 16, seconds: 71.468, TFLOP/s/device: 0.190, Tokens/s/device: 28.656, total_weights: 65536, loss: 5.599 I0417 02:21:27.312411 140462700377920 metric_logger.py:185] completed step: 17, seconds: 0.008, TFLOP/s/device: 1608.515, Tokens/s/device: 242452.942, total_weights: 65536, loss: 5.511 I0417 02:21:27.460409 140462700377920 metric_logger.py:185] completed step: 18, seconds: 0.146, TFLOP/s/device: 92.971, Tokens/s/device: 14013.672, total_weights: 65536, loss: 5.442 I0417 02:21:27.608847 140462700377920 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 02:21:27.609574 140462700377920 checkpointing.py:798] Waited 0.000751495361328125 seconds for step 19 to finish before starting checkpointing. I0417 02:21:27.611497 140462700377920 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:21:49.763248 140336392296192 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.308074s (commit=39.986188s, array_metadata_write=1.321886s) I0417 02:21:49.764493 140335883732736 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.515 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.47644925117493 s) (per-host) I0417 02:21:49.764618 140335883732736 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.420107s. I0417 02:21:51.698178 140335883732736 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000406s. I0417 02:21:51.893336 140335883732736 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:21:52.849909 140335883732736 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.199395s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15/items I0417 02:21:52.851410 140335883732736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15/items I0417 02:21:53.804795 140335883732736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15 I0417 02:21:56.082135 140335883732736 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15`. I0417 02:21:56.082879 140335883732736 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.51 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/15 I0417 02:21:56.084232 140335883732736 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.739720s. I0417 02:21:56.084445 140335900518144 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:21:56.084576 140335900518144 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:21:56.084642 140335900518144 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:21:56.086663 140335900518144 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 02:21:56.086828 140462700377920 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0417 02:21:56.086980 140462700377920 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.475488 seconds. If this number is high, consider checkpointing less frequently. I0417 02:21:56.088465 140462700377920 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0417 02:21:56.090672 140462700377920 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19. I0417 02:21:56.793608 140462700377920 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:21:56.793704 140462700377920 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 I0417 02:21:56.819708 140462700377920 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027712s I0417 02:21:56.820516 140462700377920 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 42.559 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03624534606933594 s) (per-host) I0417 02:21:56.820592 140462700377920 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.036342s (batch_requests_ready=0.002575s, total_serialization_initiated=0.033050s, others=0.000717s) I0417 02:21:56.820685 140462700377920 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040448s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.040432) I0417 02:21:56.821434 140462700377920 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19. I0417 02:21:56.821771 140335900518144 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:41:56.821734 I0417 02:21:56.823676 140462700377920 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:21:56.823924 140337485952768 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:21:56.824063 140462700377920 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392487.611462, 'wait_for_prev_duration_secs': 28.475488424301147, 'time_between_consecutive_saves_sec': 23.041420221328735, 'checkpointer_blocking_start_time': 1776392516.0885046, 'checkpointer_blocking_duration_secs': 0.7333817481994629, 'get_old_steps_start_time': 1776392516.8219059, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776392487.6098115, 'checkpoint_manager_blocking_duration_secs': 29.21421766281128} I0417 02:21:56.824228 140462700377920 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 02:21:56.824275 140462700377920 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:21:57.025482 140335883732736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19 I0417 02:21:59.177866 140339092379392 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19/items I0417 02:22:02.507743 140340178700032 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19/items/array_metadatas/process_0 I0417 02:22:41.925797 140339104958208 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.627818s (commit=40.383582s, array_metadata_write=1.244236s) I0417 02:22:41.926851 140335900518144 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.991 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.142573833465576 s) (per-host) I0417 02:22:41.926915 140335900518144 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.104496s. I0417 02:22:43.668808 140335900518144 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000408s. I0417 02:22:43.989511 140335900518144 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:22:44.957201 140335900518144 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.206932s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19/items I0417 02:22:44.959828 140335900518144 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19/items I0417 02:22:45.492565 140335900518144 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19 I0417 02:22:46.941168 140335900518144 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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19`. I0417 02:22:46.942014 140335900518144 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.85 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/19 I0417 02:22:46.943646 140335900518144 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.121224s. I0417 02:22:46.943794 140337485952768 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:22:46.943884 140337485952768 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:22:46.943933 140337485952768 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:22:46.945613 140337485952768 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 02:22:46.945764 140462700377920 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0417 02:22:46.945946 140462700377920 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:22:46.946800 140462700377920 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.742, Tokens/s/device: 13979.141, total_weights: 65536, loss: 5.385 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 02:22:59 UTC 2026 EXIT_CODE=0