XPK Start: Tue Apr 21 10:37:07 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-21 10:37:32.202517: 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) I0421 10:37:32.383141 135458295281472 max_utils.py:273] Attempting to initialize the jax distributed system... I0421 10:37:41.424650 135458295281472 distributed.py:149] Starting JAX distributed service on [::]:8482 I0421 10:37:41.427098 135458295281472 distributed.py:172] Connecting to JAX distributed service on mt-10-shardy-true-srfn5-slice-job-0-0.mt-10-shardy-true-srfn5:8482 I0421 10:37:42.399909 135458295281472 max_utils.py:284] Jax distributed system initialized! I0421 10:37:48.612694 135458295281472 max_utils.py:800] System Information: Jax Version: 0.9.2 I0421 10:37:48.612795 135458295281472 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0421 10:37:48.612836 135458295281472 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 I0421 10:37:48.612871 135458295281472 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 10:37:49.327952 135458295281472 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 10:37:49.443099 135458295281472 checkpointing.py:688] Setting up checkpoint logger... I0421 10:37:49.443215 135458295281472 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 10:37:49.443264 135458295281472 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0421 10:37:49.443476 135458295281472 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7b323491de80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 10:37:52.763745 135458295281472 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0421 10:37:52.764034 135458295281472 checkpoint_manager.py:708] [process=2][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1d105fddc0>}, handler_registry=None I0421 10:37:52.764305 135458295281472 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1d105fddc0>` 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`. I0421 10:37:52.764357 135458295281472 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1df01e0b00>` 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`. I0421 10:37:52.764410 135458295281472 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 0x7b1d105fddc0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1d105fddc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1df01e0b00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1df01e0b00>}). I0421 10:37:52.764766 135458295281472 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0421 10:37:52.764859 135458295281472 async_checkpointer.py:192] [process=2][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b1d10769b20> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0421 10:37:53.653336 135458295281472 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints I0421 10:37:54.078624 135458295281472 checkpoint_manager.py:929] [process=2][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=10), 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_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b1d102d60c0> I0421 10:37:54.078787 135458295281472 checkpointing.py:302] Checkpoint manager created! I0421 10:37:55.051472 135458295281472 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0421 10:37:55.051591 135458295281472 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0421 10:37:56.821480 135458295281472 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 10:37:56.821577 135458295281472 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 10:37:56.827053 135458295281472 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 10:37:56.827110 135458295281472 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 10:37:56.843947 135458295281472 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 10:37:56.844007 135458295281472 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 10:37:56.859960 135458295281472 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 10:37:56.860016 135458295281472 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 10:37:56.885302 135458295281472 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 10:37:56.885373 135458295281472 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 10:37:56.901340 135458295281472 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 10:37:56.901412 135458295281472 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 10:37:56.917295 135458295281472 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 10:37:56.917354 135458295281472 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 10:37:56.946988 135458295281472 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 10:37:56.947057 135458295281472 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 10:37:56.966969 135458295281472 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 10:37:56.967033 135458295281472 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 10:38:08.327740 135458295281472 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. I0421 10:38:08.330197 135458295281472 metric_logger.py:301] number parameters: 1.104 billion I0421 10:38:19.447149 135458295281472 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0421 10:38:19.599281 135458295281472 checkpointing.py:798] Waited 0.15211796760559082 seconds for step 0 to finish before starting checkpointing. I0421 10:38:19.601746 135458295281472 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 10:38:19.603524 135458295281472 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 0 I0421 10:38:19.604932 135458295281472 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/0. I0421 10:38:20.332372 135458295281472 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 10:38:20.333305 135458295281472 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0421 10:38:20.333361 135458295281472 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 I0421 10:38:20.609450 135458295281472 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277862s I0421 10:38:20.609631 135458295281472 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 5.371 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.287219762802124 s) (per-host) I0421 10:38:20.609686 135458295281472 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287286s (batch_requests_ready=0.003279s, total_serialization_initiated=0.283932s, others=0.000074s) I0421 10:38:20.609783 135458295281472 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291506s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.291487) I0421 10:38:20.610636 135458295281472 event_tracking.py:125] [process=2] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/0. I0421 10:38:20.610907 135330029917952 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 10:58:20.610881 I0421 10:38:20.616446 135458295281472 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 10:38:20.616734 135328832673536 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0421 10:38:20.616888 135458295281472 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776767899.6017272, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776767899.603561, 'checkpointer_blocking_duration_secs': 1.007462978363037, 'get_old_steps_start_time': 1776767900.611048, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776767899.5996926, 'checkpoint_manager_blocking_duration_secs': 1.0171589851379395} I0421 10:38:20.616994 135458295281472 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0421 10:38:20.617062 135458295281472 max_utils.py:750] Memstats: After params initialized: I0421 10:38:20.617108 135458295281472 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_8(process=2,(0,2,0,0)) I0421 10:38:20.617139 135458295281472 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_9(process=2,(1,2,0,0)) I0421 10:38:20.617165 135458295281472 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_12(process=2,(0,3,0,0)) I0421 10:38:20.617189 135458295281472 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_13(process=2,(1,3,0,0)) I0421 10:38:20.933782 135458295281472 metric_logger.py:196] completed step: 0, seconds: 11.116, TFLOP/s/device: 1.222, Tokens/s/device: 184.242, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0421 10:38:21.111486 135458295281472 metric_logger.py:196] completed step: 1, seconds: 1.485, TFLOP/s/device: 9.151, Tokens/s/device: 1379.351, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0421 10:38:21.973843 135458295281472 metric_logger.py:196] completed step: 2, seconds: 0.032, TFLOP/s/device: 429.959, Tokens/s/device: 64808.076, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28632.133 I0421 10:38:22.121505 135458295281472 metric_logger.py:196] completed step: 3, seconds: 0.852, TFLOP/s/device: 15.948, Tokens/s/device: 2403.841, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16855.789 I0421 10:38:22.417343 135458295281472 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.588, Tokens/s/device: 12448.562, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10661.594 I0421 10:38:22.425039 135458295281472 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.302, Tokens/s/device: 13912.760, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7245.978 I0421 10:38:23.980718 2796 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 10:38:27.405955 135328849458944 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/0/items/array_metadatas/process_2 I0421 10:38:44.789272 135458295281472 metric_logger.py:196] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.827, Tokens/s/device: 6907.531, total_weights: 65536, loss: 8.589, lm_loss: 8.589, perplexity: 5371.156 I0421 10:38:44.936732 135458295281472 metric_logger.py:196] completed step: 7, seconds: 22.218, TFLOP/s/device: 0.612, Tokens/s/device: 92.176, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4360.465 I0421 10:38:45.084459 135458295281472 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.865, Tokens/s/device: 13394.638, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.256 I0421 10:38:45.232833 135458295281472 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0421 10:38:45.233495 135458295281472 checkpointing.py:798] Waited 0.0006759166717529297 seconds for step 9 to finish before starting checkpointing. I0421 10:38:45.235419 135458295281472 checkpoint_manager.py:2020] [process=2][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 10:38:57.390795 135330029917952 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 42.613 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.06834697723389 s) (per-host) I0421 10:38:57.390919 135330029917952 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.779929s. I0421 10:39:06.233958 135330029917952 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 45.622949s. I0421 10:39:06.234211 135328832673536 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 10:39:06.234336 135328832673536 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0421 10:39:06.234404 135328832673536 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 10:39:06.237028 135328832673536 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 10:39:06.237156 135458295281472 checkpoint_manager.py:2032] [process=2][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0421 10:39:06.237225 135458295281472 checkpoint_manager.py:1452] Waiting for previous save to complete took 21.001834 seconds. If this number is high, consider checkpointing less frequently. I0421 10:39:06.239200 135458295281472 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 9 I0421 10:39:06.241238 135458295281472 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/9. I0421 10:39:06.965345 135458295281472 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0421 10:39:06.965453 135458295281472 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 I0421 10:39:07.001657 135458295281472 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037972s I0421 10:39:07.001805 135458295281472 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 34.096 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0452418327331543 s) (per-host) I0421 10:39:07.001854 135458295281472 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045300s (batch_requests_ready=0.002895s, total_serialization_initiated=0.042342s, others=0.000063s) I0421 10:39:07.001940 135458295281472 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049678s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.049663) I0421 10:39:07.002631 135458295281472 event_tracking.py:125] [process=2] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/9. I0421 10:39:07.002964 135328849458944 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 10:59:07.002929 I0421 10:39:07.004917 135458295281472 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 10:39:07.005241 135328832673536 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0421 10:39:07.005392 135458295281472 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776767925.2353666, 'wait_for_prev_duration_secs': 21.00183415412903, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776767946.2392378, 'checkpointer_blocking_duration_secs': 0.7638731002807617, 'get_old_steps_start_time': 1776767947.003134, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776767925.2337182, 'checkpoint_manager_blocking_duration_secs': 21.77162456512451} I0421 10:39:07.005561 135458295281472 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0421 10:39:07.005608 135458295281472 checkpoint_manager.py:2020] [process=2][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 10:39:13.456650 135304799307520 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260421_083238_10_shardy_true/checkpoints/9/items/array_metadatas/process_2 I0421 10:39:48.697989 135328849458944 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 37.842 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.74138593673706 s) (per-host) I0421 10:39:48.698112 135328849458944 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.695038s. I0421 10:39:57.451064 135328849458944 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 50.447974s. I0421 10:39:57.451429 135328832673536 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 10:39:57.451548 135328832673536 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0421 10:39:57.451596 135328832673536 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 10:39:57.453830 135328832673536 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 10:39:57.454010 135458295281472 checkpoint_manager.py:2032] [process=2][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0421 10:39:57.454177 135458295281472 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 10:39:57.455247 135458295281472 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.534, Tokens/s/device: 13797.006, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3551.438 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Tue Apr 21 10:40:08 UTC 2026 EXIT_CODE=0