XPK Start: Mon Apr 20 17:42:50 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 17:43:14.174211: 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) I0420 17:43:14.703329 136479172364096 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 17:43:23.745261 136479172364096 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 17:43:23.747567 136479172364096 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-4q77p-slice-job-0-0.mt-05-fp8-4q77p:8482 I0420 17:43:24.642539 136479172364096 max_utils.py:284] Jax distributed system initialized! I0420 17:43:30.836459 136479172364096 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 17:43:30.836558 136479172364096 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 17:43:30.836596 136479172364096 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 I0420 17:43:30.836630 136479172364096 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 17:43:31.526463 136479172364096 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 17:43:31.952405 136479172364096 checkpointing.py:688] Setting up checkpoint logger... I0420 17:43:31.952518 136479172364096 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 17:43:31.952564 136479172364096 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 17:43:31.952773 136479172364096 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 0x7c1fd10344d0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 17:43:34.924149 136479172364096 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 17:43:34.924377 136479172364096 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 0x7c0b907aec90>}, handler_registry=None I0420 17:43:34.924625 136479172364096 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c0b907aec90>` 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`. I0420 17:43:34.924674 136479172364096 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c0b907ad550>` 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`. I0420 17:43:34.924711 136479172364096 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 0x7c0b907aec90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c0b907aec90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c0b907ad550>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c0b907ad550>}). I0420 17:43:34.925034 136479172364096 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 17:43:34.925129 136479172364096 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c0bb81ca7a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 17:43:36.008934 136479172364096 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints I0420 17:43:36.454035 136479172364096 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=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_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c0b90599370> I0420 17:43:36.454253 136479172364096 checkpointing.py:302] Checkpoint manager created! I0420 17:43:37.725005 136479172364096 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 17:43:37.725139 136479172364096 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 17:43:39.897682 136479172364096 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 17:43:39.897799 136479172364096 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:43:39.897850 136479172364096 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:43:39.897894 136479172364096 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 17:43:39.897931 136479172364096 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:43:39.897966 136479172364096 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:43:39.898001 136479172364096 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:43:39.898036 136479172364096 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 17:43:39.898072 136479172364096 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:43:39.898125 136479172364096 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:43:39.898159 136479172364096 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 17:43:39.898191 136479172364096 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 17:43:40.316074 136479172364096 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:43:40.316179 136479172364096 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:43:40.321538 136479172364096 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:43:40.321598 136479172364096 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:43:40.337965 136479172364096 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:43:40.338025 136479172364096 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:43:40.353772 136479172364096 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:43:40.353830 136479172364096 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:43:40.414892 136479172364096 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:43:40.414978 136479172364096 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:43:40.430756 136479172364096 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:43:40.430818 136479172364096 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:43:40.446606 136479172364096 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:43:40.446667 136479172364096 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:43:40.476609 136479172364096 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 17:43:40.476678 136479172364096 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:43:40.536185 136479172364096 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 17:43:40.536265 136479172364096 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 17:43:54.428417 136479172364096 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. I0420 17:43:54.439174 136479172364096 metric_logger.py:301] number parameters: 1.104 billion I0420 17:44:08.439074 136479172364096 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 17:44:08.599918 136479172364096 checkpointing.py:798] Waited 0.16083049774169922 seconds for step 0 to finish before starting checkpointing. I0420 17:44:08.602396 136479172364096 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:44:08.604491 136479172364096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 17:44:08.606024 136479172364096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0. I0420 17:44:08.920268 136479172364096 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 17:44:08.950218 136479172364096 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0420 17:44:08.950303 136479172364096 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 I0420 17:44:09.354939 136479172364096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.407114s I0420 17:44:09.355760 136479172364096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 3.605 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.4279036521911621 s) (per-host) I0420 17:44:09.355829 136479172364096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.427991s (batch_requests_ready=0.005709s, total_serialization_initiated=0.421557s, others=0.000725s) I0420 17:44:09.355918 136479172364096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.431972s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.431954) I0420 17:44:09.356652 136479172364096 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0. I0420 17:44:09.356945 136350529234688 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:04:09.356910 I0420 17:44:09.370428 136479172364096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:44:09.370721 136349508171520 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:44:09.370877 136479172364096 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776707048.602378, 'wait_for_prev_duration_secs': 6.222724914550781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776707048.6045291, 'checkpointer_blocking_duration_secs': 0.752572774887085, 'get_old_steps_start_time': 1776707049.357128, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776707048.600402, 'checkpoint_manager_blocking_duration_secs': 0.7704381942749023} I0420 17:44:09.370999 136479172364096 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 17:44:09.371075 136479172364096 max_utils.py:750] Memstats: After params initialized: I0420 17:44:09.371145 136479172364096 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_0(process=0,(0,0,0,0)) I0420 17:44:09.371179 136479172364096 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_1(process=0,(1,0,0,0)) I0420 17:44:09.371207 136479172364096 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_4(process=0,(0,1,0,0)) I0420 17:44:09.371232 136479172364096 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_5(process=0,(1,1,0,0)) I0420 17:44:09.606914 136350045042432 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 17:44:09.741134 136479172364096 metric_logger.py:196] completed step: 0, seconds: 13.559, TFLOP/s/device: 1.002, Tokens/s/device: 151.038, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742 I0420 17:44:09.744822 136479172364096 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/tensorboard/' I0420 17:44:10.168829 136479172364096 metric_logger.py:196] completed step: 1, seconds: 1.296, TFLOP/s/device: 10.487, Tokens/s/device: 1580.729, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742 I0420 17:44:10.314982 136479172364096 metric_logger.py:196] completed step: 2, seconds: 0.429, TFLOP/s/device: 31.654, Tokens/s/device: 4771.212, total_weights: 65536, loss: 10.856, lm_loss: 10.856, perplexity: 51864.762 I0420 17:44:10.472452 136479172364096 metric_logger.py:196] completed step: 3, seconds: 0.017, TFLOP/s/device: 777.785, Tokens/s/device: 117236.247, total_weights: 65536, loss: 10.824, lm_loss: 10.824, perplexity: 50225.512 I0420 17:44:16.819810 136350028257024 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 17:44:16.965603 136348391667456 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776707056554043314, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776707056554043314, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 17:44:17.969003 2901 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 17:44:19.846307 136350003078912 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items/array_metadatas/process_0 I0420 17:44:35.165129 136479172364096 metric_logger.py:196] completed step: 4, seconds: 0.147, TFLOP/s/device: 92.453, Tokens/s/device: 13935.480, total_weights: 65536, loss: 10.793, lm_loss: 10.793, perplexity: 48689.156 I0420 17:44:35.178328 136479172364096 metric_logger.py:196] completed step: 5, seconds: 0.157, TFLOP/s/device: 86.333, Tokens/s/device: 13013.089, total_weights: 65536, loss: 10.763, lm_loss: 10.763, perplexity: 47229.012 I0420 17:44:35.330175 136479172364096 metric_logger.py:196] completed step: 6, seconds: 24.691, TFLOP/s/device: 0.550, Tokens/s/device: 82.946, total_weights: 65536, loss: 10.733, lm_loss: 10.733, perplexity: 45864.258 I0420 17:44:35.487403 136479172364096 metric_logger.py:196] completed step: 7, seconds: 0.010, TFLOP/s/device: 1320.036, Tokens/s/device: 198970.174, total_weights: 65536, loss: 10.706, lm_loss: 10.706, perplexity: 44620.242 I0420 17:44:35.644438 136479172364096 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.671, Tokens/s/device: 13365.529, total_weights: 65536, loss: 10.680, lm_loss: 10.680, perplexity: 43477.055 I0420 17:44:35.654953 136479172364096 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 17:44:35.958835 136479172364096 checkpointing.py:798] Waited 0.3038492202758789 seconds for step 10 to finish before starting checkpointing. I0420 17:44:35.961190 136479172364096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:44:38.180633 136349994686208 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 20.222115s (commit=19.024780s, array_metadata_write=1.197335s) I0420 17:44:38.181986 136350529234688 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 54.002 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 29.254088163375854 s) (per-host) I0420 17:44:38.182145 136350529234688 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.825087s. I0420 17:44:38.853533 136350529234688 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776707056554043314, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 17:44:39.610370 136348391667456 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776707056554043314, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 17:44:39.656088 136350529234688 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000901s. I0420 17:44:40.265698 136350529234688 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 17:44:41.211133 136350529234688 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.230184s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 17:44:41.213136 136350529234688 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 17:44:41.788503 136350529234688 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 17:44:43.238213 136350529234688 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0`. I0420 17:44:43.238940 136350529234688 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 34.63 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 17:44:43.240493 136350529234688 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 33.883434s. I0420 17:44:43.240695 136349508171520 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:44:43.240823 136349508171520 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:44:43.240887 136349508171520 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:44:43.243343 136349508171520 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 17:44:43.243565 136479172364096 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 17:44:43.243707 136479172364096 checkpoint_manager.py:1452] Waiting for previous save to complete took 7.282515 seconds. If this number is high, consider checkpointing less frequently. I0420 17:44:43.245469 136479172364096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0420 17:44:43.247599 136479172364096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10. I0420 17:44:43.561223 136479172364096 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0420 17:44:43.561322 136479172364096 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 I0420 17:44:43.600728 136479172364096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041980s I0420 17:44:43.601585 136479172364096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 26.002 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05933237075805664 s) (per-host) I0420 17:44:43.601658 136479172364096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.059423s (batch_requests_ready=0.005239s, total_serialization_initiated=0.053408s, others=0.000777s) I0420 17:44:43.601751 136479172364096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.063484s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.063467) I0420 17:44:43.602498 136479172364096 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10. I0420 17:44:43.602790 136349508171520 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:04:43.602756 I0420 17:44:43.813980 136350529234688 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 17:44:44.054435 136479172364096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:44:44.054826 136347863987968 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:44:44.055006 136479172364096 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776707075.9611573, 'wait_for_prev_duration_secs': 7.282514810562134, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776707083.2455087, 'checkpointer_blocking_duration_secs': 0.3574209213256836, 'get_old_steps_start_time': 1776707083.6029527, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776707075.9592998, 'checkpoint_manager_blocking_duration_secs': 8.095669746398926} I0420 17:44:44.055145 136479172364096 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 17:44:44.055941 136479172364096 metric_logger.py:196] completed step: 9, seconds: 0.157, TFLOP/s/device: 86.344, Tokens/s/device: 13014.743, total_weights: 65536, loss: 10.656, lm_loss: 10.656, perplexity: 42467.320 I0420 17:44:44.070149 136479172364096 metric_logger.py:196] completed step: 10, seconds: 0.157, TFLOP/s/device: 86.453, Tokens/s/device: 13031.140, total_weights: 65536, loss: 10.636, lm_loss: 10.636, perplexity: 41601.941 I0420 17:44:44.220569 136479172364096 metric_logger.py:196] completed step: 11, seconds: 8.412, TFLOP/s/device: 1.615, Tokens/s/device: 243.453, total_weights: 65536, loss: 10.618, lm_loss: 10.618, perplexity: 40849.570 I0420 17:44:44.377414 136479172364096 metric_logger.py:196] completed step: 12, seconds: 0.012, TFLOP/s/device: 1120.034, Tokens/s/device: 168823.675, total_weights: 65536, loss: 10.602, lm_loss: 10.602, perplexity: 40203.926 I0420 17:44:44.534330 136479172364096 metric_logger.py:196] completed step: 13, seconds: 0.152, TFLOP/s/device: 89.470, Tokens/s/device: 13485.928, total_weights: 65536, loss: 10.588, lm_loss: 10.588, perplexity: 39652.145 I0420 17:44:45.023483 136479172364096 metric_logger.py:196] completed step: 14, seconds: 0.157, TFLOP/s/device: 86.634, Tokens/s/device: 13058.393, total_weights: 65536, loss: 10.577, lm_loss: 10.577, perplexity: 39212.879 I0420 17:44:45.175695 136479172364096 metric_logger.py:196] completed step: 15, seconds: 0.634, TFLOP/s/device: 21.423, Tokens/s/device: 3229.102, total_weights: 65536, loss: 10.568, lm_loss: 10.568, perplexity: 38879.223 I0420 17:44:45.332530 136479172364096 metric_logger.py:196] completed step: 16, seconds: 0.010, TFLOP/s/device: 1321.962, Tokens/s/device: 199260.557, total_weights: 65536, loss: 10.562, lm_loss: 10.562, perplexity: 38619.105 I0420 17:44:45.489179 136479172364096 metric_logger.py:196] completed step: 17, seconds: 0.154, TFLOP/s/device: 88.502, Tokens/s/device: 13339.934, total_weights: 65536, loss: 10.556, lm_loss: 10.556, perplexity: 38395.816 I0420 17:44:45.645998 136479172364096 metric_logger.py:196] completed step: 18, seconds: 0.157, TFLOP/s/device: 86.581, Tokens/s/device: 13050.488, total_weights: 65536, loss: 10.552, lm_loss: 10.552, perplexity: 38236.652 I0420 17:44:45.804555 136479172364096 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 17:44:45.805828 136479172364096 checkpointing.py:798] Waited 0.001285552978515625 seconds for step 19 to finish before starting checkpointing. I0420 17:44:45.807854 136479172364096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:44:46.062727 136345179633408 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 17:44:49.028527 136347889166080 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items/array_metadatas/process_0 I0420 17:45:08.550855 136347872380672 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 21.337615s (commit=19.279454s, array_metadata_write=2.058161s) I0420 17:45:08.552166 136349508171520 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 63.166 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 25.00988531112671 s) (per-host) I0420 17:45:08.552296 136349508171520 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.949390s. I0420 17:45:10.890024 136349508171520 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000851s. I0420 17:45:11.085925 136349508171520 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 17:45:12.028228 136349508171520 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.220748s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 17:45:12.030924 136349508171520 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 17:45:12.590455 136349508171520 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 17:45:14.453467 136349508171520 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10`. I0420 17:45:14.454309 136349508171520 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 31.21 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 17:45:14.455667 136349508171520 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 30.852759s. I0420 17:45:14.455890 136347863987968 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:45:14.456010 136347863987968 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:45:14.456073 136347863987968 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:45:14.458950 136347863987968 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 17:45:14.459104 136479172364096 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. W0420 17:45:14.459176 136479172364096 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.651332 seconds. If this number is high, consider checkpointing less frequently. I0420 17:45:14.460883 136479172364096 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0420 17:45:14.463129 136479172364096 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19. I0420 17:45:14.781446 136479172364096 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array. I0420 17:45:14.781604 136479172364096 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 I0420 17:45:14.826420 136479172364096 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.047363s I0420 17:45:14.827240 136479172364096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 23.969 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06436324119567871 s) (per-host) I0420 17:45:14.827310 136479172364096 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.064452s (batch_requests_ready=0.005242s, total_serialization_initiated=0.058468s, others=0.000742s) I0420 17:45:14.827399 136479172364096 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.068450s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.068439) I0420 17:45:14.828119 136479172364096 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.37 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19. I0420 17:45:14.828453 136347889166080 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:05:14.828412 I0420 17:45:14.830496 136479172364096 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:45:14.830748 136347863987968 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:45:14.830914 136479172364096 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776707085.807824, 'wait_for_prev_duration_secs': 28.65133237838745, 'time_between_consecutive_saves_sec': 2.564438819885254, 'checkpointer_blocking_start_time': 1776707114.460928, 'checkpointer_blocking_duration_secs': 0.36766767501831055, 'get_old_steps_start_time': 1776707114.8286128, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776707085.806168, 'checkpoint_manager_blocking_duration_secs': 29.02471089363098} I0420 17:45:14.831046 136479172364096 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 17:45:14.831107 136479172364096 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:45:15.009849 136349508171520 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 17:45:17.053944 136355363395328 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 17:45:20.017228 136350019864320 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items/array_metadatas/process_0 I0420 17:45:39.760784 136350011471616 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 21.558818s (commit=19.863743s, array_metadata_write=1.695075s) I0420 17:45:39.762179 136347889166080 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 63.193 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 24.999273538589478 s) (per-host) I0420 17:45:39.762312 136347889166080 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.933739s. I0420 17:45:41.612787 136347889166080 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000850s. I0420 17:45:41.853371 136347889166080 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 17:45:42.778967 136347889166080 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.184787s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 17:45:42.780753 136347889166080 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 17:45:43.341471 136347889166080 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 17:45:44.815894 136347889166080 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19`. I0420 17:45:44.816889 136347889166080 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 30.36 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 17:45:44.818480 136347889166080 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 29.989906s. I0420 17:45:44.818703 136347863987968 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:45:44.818822 136347863987968 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:45:44.818873 136347863987968 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:45:44.821405 136347863987968 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 17:45:44.821549 136479172364096 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. I0420 17:45:44.821683 136479172364096 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:45:44.822632 136479172364096 metric_logger.py:196] completed step: 19, seconds: 0.157, TFLOP/s/device: 86.674, Tokens/s/device: 13064.474, total_weights: 65536, loss: 10.549, lm_loss: 10.549, perplexity: 38125.074 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 17:45:56 UTC 2026 EXIT_CODE=0