XPK Start: Fri Apr 17 02:29:08 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 02:29:32.263681: 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 02:29:32.441725 132738677065536 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:29:41.480789 132738677065536 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:29:41.483165 132738677065536 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-13abe-slice-job-0-0.mt-16-zarr3-false-rs-13abe:8482 I0417 02:29:42.647340 132738677065536 max_utils.py:284] Jax distributed system initialized! I0417 02:29:48.781918 132738677065536 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:29:48.782020 132738677065536 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:29:48.782059 132738677065536 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 02:29:48.782093 132738677065536 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:29:49.447949 132738677065536 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:29:49.560458 132738677065536 checkpointing.py:688] Setting up checkpoint logger... I0417 02:29:49.560573 132738677065536 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0417 02:29:49.560620 132738677065536 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:29:49.560821 132738677065536 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x78b8feb14740>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:29:51.236270 132738677065536 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:29:51.236505 132738677065536 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 0x78a50c278c80>}, handler_registry=None I0417 02:29:51.236752 132738677065536 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78a50c278c80>` 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 02:29:51.236800 132738677065536 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a4e44d7fe0>` 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 02:29:51.236851 132738677065536 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 0x78a50c278c80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78a50c278c80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a4e44d7fe0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a4e44d7fe0>}). I0417 02:29:51.237184 132738677065536 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:29:51.237258 132738677065536 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78a50c686020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:29:53.989524 132618355980032 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392771703073984, 'commit_timestamp_nsecs': 1776392818706183666, '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_false/checkpoints/5/_CHECKPOINT_METADATA I0417 02:29:54.236535 132618355980032 checkpoint_manager.py:1871] Missing metrics for step 5 I0417 02:29:55.089879 132738677065536 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_false/checkpoints I0417 02:29:55.092348 132738677065536 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_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78a50c27b500> I0417 02:29:55.092460 132738677065536 checkpointing.py:302] Checkpoint manager created! I0417 02:29:56.044899 132738677065536 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:29:56.045011 132738677065536 checkpointing.py:582] restoring from this run's directory step 9 I0417 02:29:56.295366 132738677065536 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_false/checkpoints/9. I0417 02:29:56.559114 132738677065536 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_false/checkpoints/9. I0417 02:29:56.559530 132738677065536 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.26 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_false/checkpoints/9. I0417 02:29:57.979099 1438 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:30:04.118033 132738677065536 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 257.134 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.145740509033203 s) (per-host) I0417 02:30:04.126485 132738677065536 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 240.760 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.573380470275879 s) (per-host) I0417 02:30:07.565480 132738677065536 event_tracking.py:138] [process=0] [sync] Finished load in 11.27 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_false/checkpoints/9 I0417 02:30:07.565634 132738677065536 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_false/checkpoints', 'checkpointer_start_time': 1776392996.2952685, 'checkpointer_duration_secs': 11.270333766937256, 'checkpoint_manager_start_time': 1776392996.045612, 'checkpoint_manager_duration_secs': 11.519991636276245} fsdp: 32 I0417 02:30:07.640931 132738677065536 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:30:07.641041 132738677065536 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:30:07.641095 132738677065536 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:30:07.641137 132738677065536 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:30:07.641174 132738677065536 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:30:07.641209 132738677065536 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:30:07.641243 132738677065536 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:30:07.641277 132738677065536 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:30:07.641310 132738677065536 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:30:07.641341 132738677065536 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:30:07.641373 132738677065536 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:30:07.641404 132738677065536 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:30:08.043219 132738677065536 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:30:08.043309 132738677065536 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:30:08.049614 132738677065536 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:30:08.049670 132738677065536 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:30:08.065903 132738677065536 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:30:08.065960 132738677065536 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:30:08.087319 132738677065536 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:30:08.087418 132738677065536 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:30:08.111677 132738677065536 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 02:30:08.111748 132738677065536 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:30:08.127445 132738677065536 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 02:30:08.127503 132738677065536 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:30:08.143070 132738677065536 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 02:30:08.143132 132738677065536 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:30:08.172203 132738677065536 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:30:08.172272 132738677065536 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:30:08.191853 132738677065536 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:30:08.191915 132738677065536 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:30:19.393857 132738677065536 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 02:30:19.402976 132738677065536 metric_logger.py:289] number parameters: 1.104 billion I0417 02:30:30.716155 132738677065536 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 02:30:30.867897 132738677065536 checkpointing.py:798] Waited 0.15173673629760742 seconds for step 10 to finish before starting checkpointing. I0417 02:30:30.870698 132738677065536 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:30:30.872657 132738677065536 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0417 02:30:30.874075 132738677065536 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_false/checkpoints/10. I0417 02:30:31.577008 132738677065536 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:30:31.594261 132738677065536 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:30:31.594382 132738677065536 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 02:30:31.830794 132616657286912 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_false/checkpoints/10 I0417 02:30:31.867473 132738677065536 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274676s I0417 02:30:31.868220 132738677065536 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.444 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2833290100097656 s) (per-host) I0417 02:30:31.868293 132738677065536 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283422s (batch_requests_ready=0.002777s, total_serialization_initiated=0.279972s, others=0.000673s) I0417 02:30:31.868376 132738677065536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287381s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.287364) I0417 02:30:31.869131 132738677065536 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.00 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_false/checkpoints/10. I0417 02:30:31.869488 132617743607552 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:50:31.869450 I0417 02:30:31.877459 132738677065536 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:30:31.877734 132613453203200 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:30:31.877902 132738677065536 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776393030.8706796, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776393030.872696, 'checkpointer_blocking_duration_secs': 0.9969456195831299, 'get_old_steps_start_time': 1776393031.8696637, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776393030.8682458, 'checkpoint_manager_blocking_duration_secs': 1.009613275527954} I0417 02:30:31.878016 132738677065536 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 02:30:31.878086 132738677065536 max_utils.py:750] Memstats: After params initialized: I0417 02:30:31.878136 132738677065536 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 02:30:31.878167 132738677065536 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 02:30:31.878190 132738677065536 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 02:30:31.878212 132738677065536 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 02:30:32.219487 132738677065536 metric_logger.py:185] completed step: 10, seconds: 11.012, TFLOP/s/device: 1.234, Tokens/s/device: 185.983, total_weights: 65536, loss: 6.827 I0417 02:30:32.360938 132738677065536 metric_logger.py:185] completed step: 11, seconds: 1.500, TFLOP/s/device: 9.058, Tokens/s/device: 1365.246, total_weights: 65536, loss: 6.522 I0417 02:30:32.508643 132738677065536 metric_logger.py:185] completed step: 12, seconds: 0.013, TFLOP/s/device: 1052.531, Tokens/s/device: 158649.005, total_weights: 65536, loss: 6.260 I0417 02:30:32.656226 132738677065536 metric_logger.py:185] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.591, Tokens/s/device: 14408.533, total_weights: 65536, loss: 6.039 I0417 02:30:32.668679 132738677065536 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 02:30:32.951203 132738677065536 checkpointing.py:798] Waited 0.28249382972717285 seconds for step 15 to finish before starting checkpointing. I0417 02:30:32.953996 132738677065536 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:30:34.028870 132614513989376 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_false/checkpoints/10/items I0417 02:30:34.199589 132613444810496 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776393033765855388, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776393033765855388, "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_false/checkpoints/10/_CHECKPOINT_METADATA I0417 02:30:37.808182 132613469988608 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_false/checkpoints/10/items/array_metadatas/process_0 I0417 02:31:16.456053 132613461595904 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.863329s (commit=39.217374s, array_metadata_write=1.645955s) I0417 02:31:16.457518 132617743607552 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.202 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.87259387969971 s) (per-host) I0417 02:31:16.457677 132617743607552 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.588073s. I0417 02:31:17.946908 132617743607552 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000749s. I0417 02:31:17.973359 132613444810496 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776393033765855388, '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_false/checkpoints/10/_CHECKPOINT_METADATA I0417 02:31:18.899034 132617743607552 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.568162s. use_zarr3=False, 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_false/checkpoints/10/items I0417 02:31:18.901235 132617743607552 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_false/checkpoints/10/items I0417 02:31:19.442223 132617743607552 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_false/checkpoints/10 I0417 02:31:21.315025 132617743607552 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_false/checkpoints/10`. I0417 02:31:21.316203 132617743607552 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.44 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_false/checkpoints/10 I0417 02:31:21.317724 132617743607552 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.448116s. I0417 02:31:21.318018 132613453203200 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:31:21.318180 132613453203200 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:31:21.318290 132613453203200 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:31:21.319975 132613453203200 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 02:31:21.320232 132738677065536 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 02:31:21.320407 132738677065536 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.366417 seconds. If this number is high, consider checkpointing less frequently. I0417 02:31:21.322377 132738677065536 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0417 02:31:21.324694 132738677065536 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_false/checkpoints/15. I0417 02:31:21.649710 132738677065536 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:31:21.649855 132738677065536 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 02:31:21.680770 132738677065536 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.033980s I0417 02:31:21.681736 132738677065536 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 28.840 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05348706245422363 s) (per-host) I0417 02:31:21.681851 132738677065536 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.053609s (batch_requests_ready=0.006772s, total_serialization_initiated=0.045955s, others=0.000882s) I0417 02:31:21.681978 132738677065536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.058374s (all_items=0.000021s, per_item={'items': '0.00002098'}, temp_paths=0.058353) I0417 02:31:21.682965 132738677065536 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.36 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_false/checkpoints/15. I0417 02:31:21.683353 132613453203200 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:51:21.683316 I0417 02:31:21.685809 132738677065536 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:31:21.686149 132613461595904 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:31:21.686302 132738677065536 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776393032.953948, 'wait_for_prev_duration_secs': 48.366416692733765, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776393081.322445, 'checkpointer_blocking_duration_secs': 0.3610873222351074, 'get_old_steps_start_time': 1776393081.6835623, 'get_old_steps_duration_secs': 4.291534423828125e-05, 'checkpoint_manager_blocking_start_time': 1776393032.9516833, 'checkpoint_manager_blocking_duration_secs': 48.7345654964447} I0417 02:31:21.686526 132738677065536 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 02:31:21.861207 132617743607552 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_false/checkpoints/15 I0417 02:31:24.934794 132613478381312 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_false/checkpoints/15/items I0417 02:31:27.807460 132614513989376 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_false/checkpoints/15/items/array_metadatas/process_0 I0417 02:31:42.717440 132738677065536 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.962, Tokens/s/device: 13861.440, total_weights: 65536, loss: 5.858 I0417 02:31:42.728033 132738677065536 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.788, Tokens/s/device: 13835.220, total_weights: 65536, loss: 5.713 I0417 02:31:42.872039 132738677065536 metric_logger.py:185] completed step: 16, seconds: 70.059, TFLOP/s/device: 0.194, Tokens/s/device: 29.232, total_weights: 65536, loss: 5.599 I0417 02:31:43.019489 132738677065536 metric_logger.py:185] completed step: 17, seconds: 0.008, TFLOP/s/device: 1615.976, Tokens/s/device: 243577.545, total_weights: 65536, loss: 5.511 I0417 02:31:43.166995 132738677065536 metric_logger.py:185] completed step: 18, seconds: 0.146, TFLOP/s/device: 92.972, Tokens/s/device: 14013.767, total_weights: 65536, loss: 5.442 I0417 02:31:43.315613 132738677065536 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 02:31:43.316339 132738677065536 checkpointing.py:798] Waited 0.0007369518280029297 seconds for step 19 to finish before starting checkpointing. I0417 02:31:43.318380 132738677065536 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:32:06.258183 132613469988608 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.212505s (commit=38.577451s, array_metadata_write=1.635054s) I0417 02:32:06.259395 132613453203200 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.392 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.63112735748291 s) (per-host) I0417 02:32:06.259523 132613453203200 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.576028s. I0417 02:32:08.507735 132613453203200 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000417s. I0417 02:32:09.103061 132613453203200 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.451809s. use_zarr3=False, 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_false/checkpoints/15/items I0417 02:32:09.104811 132613453203200 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_false/checkpoints/15/items I0417 02:32:09.641991 132613453203200 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_false/checkpoints/15 I0417 02:32:11.041214 132613453203200 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_false/checkpoints/15`. I0417 02:32:11.042037 132613453203200 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.72 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_false/checkpoints/15 I0417 02:32:11.043741 132613453203200 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.360244s. I0417 02:32:11.044026 132613461595904 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:32:11.044147 132613461595904 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:32:11.044214 132613461595904 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:32:11.052347 132613461595904 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 02:32:11.052468 132738677065536 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 02:32:11.052538 132738677065536 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.734169 seconds. If this number is high, consider checkpointing less frequently. I0417 02:32:11.054120 132738677065536 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0417 02:32:11.056190 132738677065536 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_false/checkpoints/19. I0417 02:32:11.776674 132738677065536 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:32:11.776773 132738677065536 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 02:32:11.804582 132738677065536 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029382s I0417 02:32:11.805415 132738677065536 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.238 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03833651542663574 s) (per-host) I0417 02:32:11.805487 132738677065536 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038427s (batch_requests_ready=0.002779s, total_serialization_initiated=0.034898s, others=0.000751s) I0417 02:32:11.805572 132738677065536 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042451s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.042434) I0417 02:32:11.806247 132738677065536 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 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_false/checkpoints/19. I0417 02:32:11.806566 132613461595904 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:52:11.806530 I0417 02:32:11.808507 132738677065536 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:32:11.808796 132616661473024 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:32:11.808972 132738677065536 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_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776393103.3183458, 'wait_for_prev_duration_secs': 27.734169244766235, 'time_between_consecutive_saves_sec': 21.998304843902588, 'checkpointer_blocking_start_time': 1776393131.0541594, 'checkpointer_blocking_duration_secs': 0.7525339126586914, 'get_old_steps_start_time': 1776393131.806711, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776393103.3165746, 'checkpoint_manager_blocking_duration_secs': 28.492364406585693} I0417 02:32:11.809091 132738677065536 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 02:32:11.809133 132738677065536 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:32:12.009227 132613453203200 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_false/checkpoints/19 I0417 02:32:13.749496 132616703436544 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_false/checkpoints/19/items I0417 02:32:17.069287 132617743607552 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_false/checkpoints/19/items/array_metadatas/process_0 I0417 02:32:55.046683 132616669865728 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.194367s (commit=38.975192s, array_metadata_write=1.219176s) I0417 02:32:55.047755 132613461595904 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 36.497 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.28064823150635 s) (per-host) I0417 02:32:55.047879 132613461595904 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.241208s. I0417 02:32:56.946734 132613461595904 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000399s. I0417 02:32:57.313129 132613461595904 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.445266s. use_zarr3=False, 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_false/checkpoints/19/items I0417 02:32:57.314922 132613461595904 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_false/checkpoints/19/items I0417 02:32:57.833784 132613461595904 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_false/checkpoints/19 I0417 02:32:59.537675 132613461595904 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_false/checkpoints/19`. I0417 02:32:59.538544 132613461595904 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 48.48 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_false/checkpoints/19 I0417 02:32:59.540103 132613461595904 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 47.733430s. I0417 02:32:59.540319 132616661473024 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:32:59.540430 132616661473024 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:32:59.540483 132616661473024 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:32:59.543367 132616661473024 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 02:32:59.543507 132738677065536 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 02:32:59.543623 132738677065536 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:32:59.544290 132738677065536 metric_logger.py:185] completed step: 19, seconds: 0.146, TFLOP/s/device: 92.807, Tokens/s/device: 13988.880, 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 02:33:08 UTC 2026 EXIT_CODE=0