XPK Start: Sun Apr 19 06:30:38 UTC 2026 2026-04-19 06:30:42.384818: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776580242.397694 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776580242.401375 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776580242.412649 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776580242.412669 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776580242.412671 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776580242.412673 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-19 06:31:01.430804: 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 06:31:01.960524 137231503734592 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-19 06:31:11,001:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0419 06:31:11.001769 137231503734592 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-19 06:31:11,004:jax._src.distributed:157: Connecting to JAX distributed service on mt-16-zarr3-false-rs-wul8b-slice-job-0-0.mt-16-zarr3-false-rs-wul8b:8482 I0419 06:31:11.004098 137231503734592 distributed.py:157] Connecting to JAX distributed service on mt-16-zarr3-false-rs-wul8b-slice-job-0-0.mt-16-zarr3-false-rs-wul8b:8482 I0419 06:31:12.922302 137231503734592 max_utils.py:284] Jax distributed system initialized! I0419 06:31:19.785609 137231503734592 max_utils.py:800] System Information: Jax Version: 0.8.1 I0419 06:31:19.785716 137231503734592 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0419 06:31:19.785758 137231503734592 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0419 06:31:19.785793 137231503734592 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 06:31:20.391984 137231503734592 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 06:31:20.506437 137231503734592 checkpointing.py:688] Setting up checkpoint logger... I0419 06:31:20.506553 137231503734592 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0419 06:31:20.506597 137231503734592 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0419 06:31:20.506798 137231503734592 base_pytree_checkpoint_handler.py:415] 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 0x7ccf36b79160>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 06:31:23.488739 137231503734592 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 06:31:23.489100 137231503734592 checkpoint_manager.py:709] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cbaa42a0530>}, handler_registry=None I0419 06:31:23.489338 137231503734592 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cbaa42a0530>` 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 06:31:23.489390 137231503734592 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ccdedd1bfe0>` 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 06:31:23.489426 137231503734592 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 0x7cbaa42a0530>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cbaa42a0530>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ccdedd1bfe0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ccdedd1bfe0>}). I0419 06:31:23.489767 137231503734592 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0419 06:31:23.489858 137231503734592 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7cbbd3dacf40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0419 06:31:24.932842 137110874986240 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776580062285417131, 'commit_timestamp_nsecs': 1776580110691785136, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/5/_CHECKPOINT_METADATA I0419 06:31:25.187038 137110874986240 checkpoint_manager.py:1877] Missing metrics for step 5 I0419 06:31:25.604779 137231503734592 checkpoint_manager.py:1818] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints I0419 06:31:27.914866 137231503734592 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_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7cbbd3d47740> I0419 06:31:27.915040 137231503734592 checkpointing.py:302] Checkpoint manager created! I0419 06:31:28.825347 137231503734592 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 06:31:28.825467 137231503734592 checkpointing.py:582] restoring from this run's directory step 9 I0419 06:31:29.747469 137231503734592 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/9. I0419 06:31:31.825216 1425 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 06:31:38.673086 137231503734592 jax_array_handlers.py:837] [process=0] /jax/orbax/read/worker/io/requested throughput: 230.580 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.853470325469971 s) (per-host) I0419 06:31:38.682301 137231503734592 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/read/gbytes_per_sec: 217.962 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.260934829711914 s) (per-host) I0419 06:31:40.304064 137231503734592 checkpointer.py:318] Finished restoring checkpoint in 10.80 seconds from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/9. I0419 06:31:40.304254 137231503734592 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776580289.5055356, 'checkpointer_duration_secs': 10.798685312271118, 'checkpoint_manager_start_time': 1776580288.8260722, 'checkpoint_manager_duration_secs': 11.478151082992554} fsdp: 32 I0419 06:31:40.385123 137231503734592 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 06:31:40.385241 137231503734592 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 06:31:40.385294 137231503734592 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 06:31:40.385338 137231503734592 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 06:31:40.385378 137231503734592 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 06:31:40.385417 137231503734592 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 06:31:40.385460 137231503734592 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 06:31:40.385498 137231503734592 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 06:31:40.385535 137231503734592 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 06:31:40.385568 137231503734592 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 06:31:40.385601 137231503734592 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 06:31:40.385636 137231503734592 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 06:31:40.782620 137231503734592 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 06:31:40.782770 137231503734592 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 06:31:40.789047 137231503734592 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 06:31:40.789196 137231503734592 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 06:31:40.944318 137231503734592 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 06:31:40.944451 137231503734592 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 06:31:41.096057 137231503734592 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 06:31:41.096187 137231503734592 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 06:31:41.260033 137231503734592 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 06:31:41.260172 137231503734592 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 06:31:41.414848 137231503734592 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 06:31:41.414978 137231503734592 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 06:31:41.571045 137231503734592 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 06:31:41.571171 137231503734592 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 06:31:41.736204 137231503734592 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 06:31:41.736333 137231503734592 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 06:31:41.758978 137231503734592 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 06:31:41.759070 137231503734592 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 06:31:52.827432 137231503734592 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 06:31:52.834908 137231503734592 metric_logger.py:289] number parameters: 1.104 billion I0419 06:32:04.365746 137231503734592 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0419 06:32:04.517698 137231503734592 checkpointing.py:798] Waited 0.15193557739257812 seconds for step 10 to finish before starting checkpointing. I0419 06:32:04.520271 137231503734592 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 06:32:04.521993 137231503734592 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 10 I0419 06:32:04.524722 137231503734592 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10. I0419 06:32:05.292336 137231503734592 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 06:32:05.588862 137105875392256 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10 I0419 06:32:05.589824 137231503734592 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0419 06:32:05.589943 137231503734592 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 06:32:05.835451 137231503734592 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.247833s I0419 06:32:05.836353 137231503734592 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.878 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5359632968902588 s) (per-host) I0419 06:32:05.836477 137231503734592 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.536047s (batch_requests_ready=0.281122s, total_serialization_initiated=0.254108s, others=0.000817s) I0419 06:32:05.836588 137231503734592 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.540397s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.540378) I0419 06:32:05.837448 137107579066112 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0419 06:32:05.837604 137231503734592 async_checkpointer.py:561] Finished blocking save. Time taken: 1.315539s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10. I0419 06:32:06.267211 137231503734592 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0419 06:32:06.267578 137105866999552 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 06:32:06.267740 137231503734592 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776580324.5202532, 'wait_for_prev_duration_secs': 6.437301635742188e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776580324.5220327, 'checkpointer_blocking_duration_secs': 1.3156795501708984, 'get_old_steps_start_time': 1776580325.8377314, 'get_old_steps_duration_secs': 2.5987625122070312e-05, 'checkpoint_manager_blocking_start_time': 1776580324.5181034, 'checkpoint_manager_blocking_duration_secs': 1.7495980262756348} I0419 06:32:06.267863 137231503734592 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0419 06:32:06.267943 137231503734592 max_utils.py:750] Memstats: After params initialized: I0419 06:32:06.268002 137231503734592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0419 06:32:06.268037 137231503734592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0419 06:32:06.268066 137231503734592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0419 06:32:06.268102 137231503734592 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0419 06:32:06.572323 137231503734592 metric_logger.py:185] completed step: 10, seconds: 11.245, TFLOP/s/device: 1.208, Tokens/s/device: 182.120, total_weights: 65536, loss: 6.827 I0419 06:32:06.713523 137231503734592 metric_logger.py:185] completed step: 11, seconds: 2.203, TFLOP/s/device: 6.167, Tokens/s/device: 929.580, total_weights: 65536, loss: 6.522 I0419 06:32:06.861217 137231503734592 metric_logger.py:185] completed step: 12, seconds: 0.013, TFLOP/s/device: 1049.604, Tokens/s/device: 158207.802, total_weights: 65536, loss: 6.260 I0419 06:32:07.008742 137231503734592 metric_logger.py:185] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.374, Tokens/s/device: 14375.864, total_weights: 65536, loss: 6.039 I0419 06:32:07.020933 137231503734592 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0419 06:32:07.303552 137231503734592 checkpointing.py:798] Waited 0.28258562088012695 seconds for step 15 to finish before starting checkpointing. I0419 06:32:07.306203 137231503734592 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 06:32:07.846490 137110304577280 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/items I0419 06:32:07.983992 137107100137216 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776580327580939625, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776580327580939625, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0419 06:32:11.953468 137108131915520 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_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/items/array_metadatas/process_0 I0419 06:32:52.192839 137108123522816 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.363152s (commit=40.796976s, array_metadata_write=1.566176s) I0419 06:32:52.194260 137107579066112 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 33.685 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.89383006095886 s) (per-host) I0419 06:32:52.194420 137107579066112 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.356859s. I0419 06:32:54.138370 137107579066112 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000753s. I0419 06:32:54.154921 137107100137216 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776580327580939625, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0419 06:32:55.225439 137107579066112 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.104144s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/items I0419 06:32:55.227639 137107579066112 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10/items I0419 06:32:55.647603 137107579066112 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10 I0419 06:32:56.953984 137107579066112 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10`. I0419 06:32:56.954918 137107579066112 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.432868s. directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/10 I0419 06:32:56.956529 137107579066112 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.118967s. I0419 06:32:56.956796 137105866999552 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 06:32:56.956999 137105866999552 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 06:32:56.957117 137105866999552 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0419 06:32:56.958949 137105866999552 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0419 06:32:56.959290 137231503734592 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0419 06:32:56.959477 137231503734592 checkpoint_manager.py:1458] Waiting for previous save to complete took 49.653280 seconds. If this number is high, consider checkpointing less frequently. I0419 06:32:56.961389 137231503734592 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 15 I0419 06:32:56.965700 137231503734592 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15. I0419 06:32:57.917750 137107579066112 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15 I0419 06:32:57.961676 137231503734592 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0419 06:32:57.961827 137231503734592 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 06:32:58.012285 137231503734592 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054797s I0419 06:32:58.013353 137231503734592 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.703 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.32797908782958984 s) (per-host) I0419 06:32:58.013465 137231503734592 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.328116s (batch_requests_ready=0.260199s, total_serialization_initiated=0.066945s, others=0.000972s) I0419 06:32:58.013639 137231503734592 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.333062s (all_items=0.000022s, per_item={'items': '0.00002217'}, temp_paths=0.333040) I0419 06:32:58.014952 137107091744512 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0419 06:32:58.015160 137231503734592 async_checkpointer.py:561] Finished blocking save. Time taken: 1.053662s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15. I0419 06:32:58.404088 137231503734592 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0419 06:32:58.404568 137105866999552 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 06:32:58.404824 137231503734592 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776580327.3061597, 'wait_for_prev_duration_secs': 49.65328049659729, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776580376.961456, 'checkpointer_blocking_duration_secs': 1.0538582801818848, 'get_old_steps_start_time': 1776580378.015337, 'get_old_steps_duration_secs': 3.910064697265625e-05, 'checkpoint_manager_blocking_start_time': 1776580327.3040502, 'checkpoint_manager_blocking_duration_secs': 51.10070276260376} I0419 06:32:58.404976 137231503734592 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0419 06:33:16.204656 137231503734592 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.387, Tokens/s/device: 13925.626, total_weights: 65536, loss: 5.858 I0419 06:33:16.215576 137231503734592 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.855, Tokens/s/device: 13845.322, total_weights: 65536, loss: 5.713 I0419 06:33:16.359590 137231503734592 metric_logger.py:185] completed step: 16, seconds: 69.194, TFLOP/s/device: 0.196, Tokens/s/device: 29.598, total_weights: 65536, loss: 5.599 I0419 06:33:16.507051 137231503734592 metric_logger.py:185] completed step: 17, seconds: 0.009, TFLOP/s/device: 1589.138, Tokens/s/device: 239532.164, total_weights: 65536, loss: 5.511 I0419 06:33:16.654822 137231503734592 metric_logger.py:185] completed step: 18, seconds: 0.146, TFLOP/s/device: 92.964, Tokens/s/device: 14012.521, total_weights: 65536, loss: 5.442 I0419 06:33:16.803246 137231503734592 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0419 06:33:16.803951 137231503734592 checkpointing.py:798] Waited 0.0007169246673583984 seconds for step 19 to finish before starting checkpointing. I0419 06:33:16.806057 137231503734592 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 06:33:17.141925 137110304577280 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15/items I0419 06:33:21.225967 137108131915520 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_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15/items/array_metadatas/process_0 I0419 06:34:01.124359 137108123522816 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.428688s (commit=40.866942s, array_metadata_write=1.561746s) I0419 06:34:01.125565 137107091744512 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 24.899 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 63.440170764923096 s) (per-host) I0419 06:34:01.125686 137107091744512 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 63.110591s. I0419 06:34:03.483498 137107091744512 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000411s. I0419 06:34:05.084586 137107091744512 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.202963s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15/items I0419 06:34:05.086619 137107091744512 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15/items I0419 06:34:05.498732 137107091744512 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15 I0419 06:34:06.818942 137107091744512 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15`. I0419 06:34:06.819705 137107091744512 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 69.858222s. directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/15 I0419 06:34:06.821272 137107091744512 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 68.806176s. I0419 06:34:06.821493 137105866999552 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 06:34:06.821624 137105866999552 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 06:34:06.821702 137105866999552 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0419 06:34:06.824390 137105866999552 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0419 06:34:06.824603 137231503734592 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0419 06:34:06.824735 137231503734592 checkpoint_manager.py:1458] Waiting for previous save to complete took 50.018694 seconds. If this number is high, consider checkpointing less frequently. I0419 06:34:06.826209 137231503734592 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 19 I0419 06:34:06.829751 137231503734592 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19. I0419 06:34:08.224095 137107091744512 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19 I0419 06:34:08.666165 137231503734592 jax_array_handlers.py:358] Scheduling D2H of 69 prioritized jax.Array. I0419 06:34:08.666262 137231503734592 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 06:34:08.694913 137231503734592 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.030919s I0419 06:34:08.695668 137231503734592 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.157 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7151026725769043 s) (per-host) I0419 06:34:08.695744 137231503734592 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.715199s (batch_requests_ready=0.677951s, total_serialization_initiated=0.036570s, others=0.000678s) I0419 06:34:08.695846 137231503734592 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.719194s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.719177) I0419 06:34:08.696684 137108123522816 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0419 06:34:08.696853 137231503734592 async_checkpointer.py:561] Finished blocking save. Time taken: 1.870576s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19. I0419 06:34:09.162324 137231503734592 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0419 06:34:09.162686 137105866999552 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 06:34:09.162858 137231503734592 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776580396.806024, 'wait_for_prev_duration_secs': 50.018693685531616, 'time_between_consecutive_saves_sec': 19.847005367279053, 'checkpointer_blocking_start_time': 1776580446.826249, 'checkpointer_blocking_duration_secs': 1.870711326599121, 'get_old_steps_start_time': 1776580448.6969743, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776580396.8041906, 'checkpoint_manager_blocking_duration_secs': 52.35861277580261} I0419 06:34:09.163040 137231503734592 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0419 06:34:09.163084 137231503734592 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 06:34:10.422605 137110296184576 atomicity.py:137] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19/items I0419 06:34:14.933191 137109734135552 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_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19/items/array_metadatas/process_0 I0419 06:34:55.708569 137108131915520 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 43.303746s (commit=40.944086s, array_metadata_write=2.359660s) I0419 06:34:55.709585 137108123522816 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 33.095 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 47.728992223739624 s) (per-host) I0419 06:34:55.709703 137108123522816 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 47.012909s. I0419 06:34:58.793107 137108123522816 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000436s. I0419 06:34:59.718891 137108123522816 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.366863s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19/items I0419 06:34:59.722828 137108123522816 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19/items I0419 06:35:00.598270 137108123522816 atomicity.py:608] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19 I0419 06:35:01.859700 137108123522816 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19`. I0419 06:35:01.860382 137108123522816 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 55.034119s. directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/nnx_xpk_feat_nnx_set_defaults_true_20260419_025828_16_zarr3_false/checkpoints/19 I0419 06:35:01.861525 137108123522816 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 53.164729s. I0419 06:35:01.861748 137105866999552 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 06:35:01.861886 137105866999552 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 06:35:01.861939 137105866999552 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0419 06:35:01.863952 137105866999552 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0419 06:35:01.864086 137231503734592 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0419 06:35:01.864197 137231503734592 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 06:35:01.864830 137231503734592 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.625, Tokens/s/device: 13961.510, 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: Sun Apr 19 06:35:14 UTC 2026 EXIT_CODE=0