XPK Start: Sun Apr 19 02:53:08 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 02:53:32.393778: 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) I0419 02:53:32.572244 138517713450816 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 02:53:41.613971 138517713450816 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 02:53:41.616301 138517713450816 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-f0kzf-slice-job-0-0.mt-16-zarr3-false-rs-f0kzf:8482 I0419 02:53:43.059672 138517713450816 max_utils.py:284] Jax distributed system initialized! I0419 02:53:49.214590 138517713450816 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 02:53:49.214695 138517713450816 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 02:53:49.214736 138517713450816 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 I0419 02:53:49.214769 138517713450816 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 02:53:50.249307 138517713450816 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 02:53:50.361258 138517713450816 checkpointing.py:688] Setting up checkpoint logger... I0419 02:53:50.361373 138517713450816 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0419 02:53:50.361421 138517713450816 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 02:53:50.361606 138517713450816 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 0x7dfa9e918770>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 02:53:52.416492 138517713450816 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 02:53:52.416727 138517713450816 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 0x7de63c21c620>}, handler_registry=None I0419 02:53:52.417022 138517713450816 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7de63c21c620>` 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`. I0419 02:53:52.417079 138517713450816 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7df9673e5cd0>` 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`. I0419 02:53:52.417116 138517713450816 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 0x7de63c21c620>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7de63c21c620>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7df9673e5cd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7df9673e5cd0>}). I0419 02:53:52.417435 138517713450816 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 02:53:52.417505 138517713450816 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7de5602e84a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 02:53:54.922640 138396345276160 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776567066482406267, 'commit_timestamp_nsecs': 1776567113125779844, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0419 02:53:55.167030 138396345276160 checkpoint_manager.py:1871] Missing metrics for step 9 I0419 02:53:56.022350 138517713450816 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints I0419 02:53:56.863893 138517713450816 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7de63c2111f0> I0419 02:53:56.864061 138517713450816 checkpointing.py:302] Checkpoint manager created! I0419 02:53:57.803729 138517713450816 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 02:53:57.803857 138517713450816 checkpointing.py:582] restoring from this run's directory step 9 I0419 02:53:58.057467 138517713450816 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/9. I0419 02:53:58.296724 138517713450816 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/9. I0419 02:53:58.297173 138517713450816 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.24 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/9. I0419 02:54:00.586508 1456 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 02:54:07.804286 138517713450816 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 218.729 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.224806547164917 s) (per-host) I0419 02:54:07.813589 138517713450816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 186.007 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 8.50832223892212 s) (per-host) I0419 02:54:08.848216 138517713450816 event_tracking.py:138] [process=0] [sync] Finished load in 10.79 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/9 I0419 02:54:08.848369 138517713450816 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776567238.0573692, 'checkpointer_duration_secs': 10.790969610214233, 'checkpoint_manager_start_time': 1776567237.8044593, 'checkpoint_manager_duration_secs': 11.043882131576538} fsdp: 32 I0419 02:54:08.923817 138517713450816 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 02:54:08.923930 138517713450816 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:54:08.923980 138517713450816 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:54:08.924021 138517713450816 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 02:54:08.924058 138517713450816 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:54:08.924093 138517713450816 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:54:08.924128 138517713450816 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:54:08.924162 138517713450816 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 02:54:08.924196 138517713450816 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:54:08.924227 138517713450816 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:54:08.924258 138517713450816 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 02:54:08.924292 138517713450816 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 02:54:09.324836 138517713450816 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:54:09.324925 138517713450816 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:54:09.330345 138517713450816 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:54:09.330400 138517713450816 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:54:09.346621 138517713450816 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:54:09.346681 138517713450816 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:54:09.362249 138517713450816 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:54:09.362305 138517713450816 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:54:09.385903 138517713450816 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:54:09.385969 138517713450816 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:54:09.401746 138517713450816 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:54:09.401819 138517713450816 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:54:09.417399 138517713450816 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:54:09.417464 138517713450816 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:54:09.446993 138517713450816 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 02:54:09.447064 138517713450816 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:54:09.466659 138517713450816 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 02:54:09.466723 138517713450816 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 02:54:20.869583 138517713450816 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. I0419 02:54:20.878963 138517713450816 metric_logger.py:301] number parameters: 1.104 billion I0419 02:54:32.560159 138517713450816 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0419 02:54:32.713198 138517713450816 checkpointing.py:798] Waited 0.15302515029907227 seconds for step 10 to finish before starting checkpointing. I0419 02:54:32.715791 138517713450816 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:54:32.717891 138517713450816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0419 02:54:32.719831 138517713450816 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10. I0419 02:54:33.456052 138517713450816 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 02:54:33.473198 138517713450816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:54:33.473320 138517713450816 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 I0419 02:54:33.718285 138391446816512 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10 I0419 02:54:33.745173 138517713450816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.273378s I0419 02:54:33.745930 138517713450816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.471 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28195619583129883 s) (per-host) I0419 02:54:33.745996 138517713450816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.282041s (batch_requests_ready=0.002788s, total_serialization_initiated=0.278573s, others=0.000680s) I0419 02:54:33.746076 138517713450816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285996s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.285979) I0419 02:54:33.746734 138517713450816 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10. I0419 02:54:33.747070 138394671761152 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 03:14:33.747037 I0419 02:54:33.754728 138517713450816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:54:33.755060 138390422943488 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:54:33.755223 138517713450816 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776567272.7157607, 'wait_for_prev_duration_secs': 0.00013184547424316406, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776567272.717931, 'checkpointer_blocking_duration_secs': 1.0292859077453613, 'get_old_steps_start_time': 1776567273.7472408, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776567272.7135577, 'checkpoint_manager_blocking_duration_secs': 1.0416300296783447} I0419 02:54:33.755328 138517713450816 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0419 02:54:33.755402 138517713450816 max_utils.py:750] Memstats: After params initialized: I0419 02:54:33.755465 138517713450816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0419 02:54:33.755505 138517713450816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0419 02:54:33.755532 138517713450816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0419 02:54:33.755558 138517713450816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0419 02:54:34.112510 138517713450816 metric_logger.py:196] completed step: 10, seconds: 11.356, TFLOP/s/device: 1.196, Tokens/s/device: 180.338, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0419 02:54:34.253643 138517713450816 metric_logger.py:196] completed step: 11, seconds: 1.549, TFLOP/s/device: 8.773, Tokens/s/device: 1322.327, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0419 02:54:34.401021 138517713450816 metric_logger.py:196] completed step: 12, seconds: 0.013, TFLOP/s/device: 1015.025, Tokens/s/device: 152995.667, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0419 02:54:34.548695 138517713450816 metric_logger.py:196] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.411, Tokens/s/device: 14381.316, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0419 02:54:34.561429 138517713450816 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0419 02:54:34.844048 138517713450816 checkpointing.py:798] Waited 0.28258395195007324 seconds for step 15 to finish before starting checkpointing. I0419 02:54:34.846791 138517713450816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:54:35.482983 138390959814400 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/items I0419 02:54:36.050938 138390414550784 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776567275213289376, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776567275213289376, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0419 02:54:39.207181 138390943028992 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/items/array_metadatas/process_0 I0419 02:55:19.993122 138390934636288 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.970638s (commit=40.743080s, array_metadata_write=2.227558s) I0419 02:55:19.994638 138394671761152 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.948 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.53062438964844 s) (per-host) I0419 02:55:19.994828 138394671761152 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.247631s. I0419 02:55:22.336086 138390414550784 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776567275213289376, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0419 02:55:23.604125 138394671761152 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000838s. I0419 02:55:24.083898 138394671761152 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.354956s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/items I0419 02:55:24.086071 138394671761152 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10/items I0419 02:55:24.619132 138394671761152 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10 I0419 02:55:26.078747 138394671761152 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10`. I0419 02:55:26.079959 138394671761152 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 53.36 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/10 I0419 02:55:26.082645 138394671761152 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.335462s. I0419 02:55:26.082962 138390422943488 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:55:26.083135 138390422943488 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:55:26.083237 138390422943488 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:55:26.084737 138390422943488 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0419 02:55:26.084979 138517713450816 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. W0419 02:55:26.085108 138517713450816 checkpoint_manager.py:1452] Waiting for previous save to complete took 51.238324 seconds. If this number is high, consider checkpointing less frequently. I0419 02:55:26.087263 138517713450816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0419 02:55:26.089767 138517713450816 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15. I0419 02:55:26.819194 138517713450816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:55:26.819326 138517713450816 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 I0419 02:55:26.850377 138517713450816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034098s I0419 02:55:26.851288 138517713450816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 29.054 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05309319496154785 s) (per-host) I0419 02:55:26.851391 138517713450816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.053219s (batch_requests_ready=0.006578s, total_serialization_initiated=0.045800s, others=0.000841s) I0419 02:55:26.851518 138517713450816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.058075s (all_items=0.000023s, per_item={'items': '0.00002313'}, temp_paths=0.058052) I0419 02:55:26.852564 138517713450816 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15. I0419 02:55:26.852919 138390422943488 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 03:15:26.852891 I0419 02:55:26.855309 138517713450816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:55:26.855627 138390943028992 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:55:26.855818 138517713450816 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776567274.8467443, 'wait_for_prev_duration_secs': 51.23832416534424, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776567326.0873322, 'checkpointer_blocking_duration_secs': 0.7657206058502197, 'get_old_steps_start_time': 1776567326.853083, 'get_old_steps_duration_secs': 4.38690185546875e-05, 'checkpoint_manager_blocking_start_time': 1776567274.8445444, 'checkpoint_manager_blocking_duration_secs': 52.01120924949646} I0419 02:55:26.855973 138517713450816 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0419 02:55:27.469103 138394671761152 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15 I0419 02:55:29.639855 138390389372672 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15/items I0419 02:55:33.374382 138390959814400 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15/items/array_metadatas/process_0 I0419 02:55:48.525734 138517713450816 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.564, Tokens/s/device: 13952.189, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0419 02:55:48.537243 138517713450816 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.654, Tokens/s/device: 13815.062, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0419 02:55:48.681108 138517713450816 metric_logger.py:196] completed step: 16, seconds: 73.975, TFLOP/s/device: 0.184, Tokens/s/device: 27.685, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0419 02:55:48.828583 138517713450816 metric_logger.py:196] completed step: 17, seconds: 0.009, TFLOP/s/device: 1475.419, Tokens/s/device: 222391.139, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0419 02:55:48.976449 138517713450816 metric_logger.py:196] completed step: 18, seconds: 0.145, TFLOP/s/device: 93.597, Tokens/s/device: 14107.987, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0419 02:55:49.124605 138517713450816 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0419 02:55:49.125317 138517713450816 checkpointing.py:798] Waited 0.0007262229919433594 seconds for step 19 to finish before starting checkpointing. I0419 02:55:49.127527 138517713450816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:56:14.014929 138390951421696 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.821816s (commit=41.210133s, array_metadata_write=1.611683s) I0419 02:56:14.015997 138390422943488 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.453 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 47.217812061309814 s) (per-host) I0419 02:56:14.016063 138390422943488 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 47.163041s. I0419 02:56:17.053561 138390422943488 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000388s. I0419 02:56:17.225690 138390422943488 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.915612s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15/items I0419 02:56:17.227294 138390422943488 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15/items I0419 02:56:17.747878 138390422943488 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15 I0419 02:56:19.144157 138390422943488 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15`. I0419 02:56:19.145002 138390422943488 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 53.06 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/15 I0419 02:56:19.146574 138390422943488 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.293548s. I0419 02:56:19.146734 138390943028992 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:56:19.146825 138390943028992 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:56:19.146897 138390943028992 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:56:19.148445 138390943028992 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0419 02:56:19.148638 138517713450816 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. W0419 02:56:19.148762 138517713450816 checkpoint_manager.py:1452] Waiting for previous save to complete took 30.021243 seconds. If this number is high, consider checkpointing less frequently. I0419 02:56:19.150161 138517713450816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0419 02:56:19.152177 138517713450816 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19. I0419 02:56:19.944162 138517713450816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:56:19.944259 138517713450816 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 I0419 02:56:19.970887 138517713450816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028188s I0419 02:56:19.971700 138517713450816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.757 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0369415283203125 s) (per-host) I0419 02:56:19.971776 138517713450816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037039s (batch_requests_ready=0.002795s, total_serialization_initiated=0.033505s, others=0.000739s) I0419 02:56:19.971888 138517713450816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041140s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.041122) I0419 02:56:19.972661 138517713450816 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.82 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19. I0419 02:56:19.972967 138390943028992 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 03:16:19.972939 I0419 02:56:19.974905 138517713450816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:56:19.975131 138391446816512 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:56:19.975244 138517713450816 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776567349.1274927, 'wait_for_prev_duration_secs': 30.02124309539795, 'time_between_consecutive_saves_sec': 23.04269003868103, 'checkpointer_blocking_start_time': 1776567379.1502018, 'checkpointer_blocking_duration_secs': 0.8228769302368164, 'get_old_steps_start_time': 1776567379.9730983, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776567349.125545, 'checkpoint_manager_blocking_duration_secs': 30.84966802597046} I0419 02:56:19.975361 138517713450816 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0419 02:56:19.975402 138517713450816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:56:20.184641 138390422943488 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19 I0419 02:56:22.379538 138394675947264 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19/items I0419 02:56:25.733563 138394717910784 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19/items/array_metadatas/process_0 I0419 02:57:07.077279 138394684339968 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 43.076382s (commit=41.472365s, array_metadata_write=1.604017s) I0419 02:57:07.078362 138390943028992 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.506 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 47.143596172332764 s) (per-host) I0419 02:57:07.078427 138390943028992 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 47.105374s. I0419 02:57:09.037646 138390943028992 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000390s. I0419 02:57:09.848207 138390943028992 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.927635s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19/items I0419 02:57:09.850014 138390943028992 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19/items I0419 02:57:10.360368 138390943028992 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19 I0419 02:57:12.237102 138390943028992 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19`. I0419 02:57:12.237912 138390943028992 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 53.09 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_16_zarr3_false/checkpoints/19 I0419 02:57:12.239488 138390943028992 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.266431s. I0419 02:57:12.239676 138391446816512 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:57:12.239791 138391446816512 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:57:12.239874 138391446816512 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:57:12.242213 138391446816512 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0419 02:57:12.242405 138517713450816 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. I0419 02:57:12.242573 138517713450816 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:57:12.243499 138517713450816 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.507, Tokens/s/device: 13943.640, total_weights: 65536, loss: 5.385, lm_loss: 5.385, perplexity: 218.045 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 02:57:24 UTC 2026 EXIT_CODE=0