XPK Start: Fri Apr 17 04:05:21 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 04:05:45.160883: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0417 04:05:45.339377 134934061111104 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 04:05:54.378318 134934061111104 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 04:05:54.380569 134934061111104 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-wpx59-slice-job-0-0.mt-15-ocdbt-true-rs-wpx59:8482 I0417 04:05:55.853303 134934061111104 max_utils.py:284] Jax distributed system initialized! I0417 04:06:02.108770 134934061111104 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 04:06:02.108894 134934061111104 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 04:06:02.108938 134934061111104 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0417 04:06:02.108975 134934061111104 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 04:06:03.140675 134934061111104 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:06:03.141281 134934061111104 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:06:03.141459 134934061111104 checkpointing.py:688] Setting up checkpoint logger... I0417 04:06:03.141511 134934061111104 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 04:06:03.141553 134934061111104 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 04:06:03.142028 134934061111104 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 0x7ab82851d0d0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 04:06:04.400653 134934061111104 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 04:06:04.400897 134934061111104 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 0x7ab7054612b0>}, handler_registry=None I0417 04:06:04.401136 134934061111104 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab7054612b0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0417 04:06:04.401188 134934061111104 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab70536d400>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0417 04:06:04.401224 134934061111104 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 0x7ab7054612b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab7054612b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab70536d400>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab70536d400>}). I0417 04:06:04.401550 134934061111104 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 04:06:04.401618 134934061111104 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ab705300ae0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 04:06:06.396878 134813117839104 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776398538921482442, 'commit_timestamp_nsecs': 1776398585928434553, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/5/_CHECKPOINT_METADATA I0417 04:06:06.639446 134813117839104 checkpoint_manager.py:1871] Missing metrics for step 5 I0417 04:06:07.486949 134934061111104 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints I0417 04:06:07.896475 134934061111104 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/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ab70536f950> I0417 04:06:07.896644 134934061111104 checkpointing.py:302] Checkpoint manager created! I0417 04:06:08.817015 134934061111104 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:06:08.817118 134934061111104 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:06:09.193381 134934061111104 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:06:09.193469 134934061111104 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:06:09.209541 134934061111104 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:06:09.209599 134934061111104 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:06:09.232622 134934061111104 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:06:09.232687 134934061111104 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:06:09.248821 134934061111104 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:06:09.248894 134934061111104 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:06:09.264994 134934061111104 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 04:06:09.265054 134934061111104 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:06:09.289260 134934061111104 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 04:06:09.289327 134934061111104 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:06:09.310920 134934061111104 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 04:06:09.310984 134934061111104 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 04:06:09.522884 134934061111104 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 04:06:09.522984 134934061111104 checkpointing.py:582] restoring from this run's directory step 9 I0417 04:06:09.774462 134934061111104 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 04:06:10.023672 134934061111104 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 04:06:10.024136 134934061111104 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 04:06:11.441312 1442 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 04:06:17.557113 134934061111104 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 258.139 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.121792316436768 s) (per-host) I0417 04:06:17.562914 134934061111104 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 241.639 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.549418687820435 s) (per-host) I0417 04:06:20.869392 134934061111104 event_tracking.py:138] [process=0] [sync] Finished load in 11.09 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9 I0417 04:06:20.869548 134934061111104 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776398769.7743661, 'checkpointer_duration_secs': 11.095150470733643, 'checkpoint_manager_start_time': 1776398769.5232368, 'checkpoint_manager_duration_secs': 11.346282243728638} fsdp: 32 I0417 04:06:21.280783 134934061111104 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 04:06:21.280920 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 04:06:21.280975 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 04:06:21.281028 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 04:06:21.281079 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 04:06:21.281126 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 04:06:21.281182 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 04:06:21.281234 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0417 04:06:21.281272 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0417 04:06:21.281307 134934061111104 maxtext_utils.py:1796] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 04:06:21.281352 134934061111104 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 04:06:21.281397 134934061111104 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 04:06:21.763544 134934061111104 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0417 04:06:21.763638 134934061111104 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0417 04:06:21.779292 134934061111104 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0417 04:06:21.779349 134934061111104 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0417 04:06:32.615387 134934061111104 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0417 04:06:32.622929 134934061111104 metric_logger.py:289] number parameters: 1.104 billion I0417 04:06:44.473321 134934061111104 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 04:06:44.627027 134934061111104 checkpointing.py:798] Waited 0.15352296829223633 seconds for step 10 to finish before starting checkpointing. I0417 04:06:44.629344 134934061111104 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:06:44.631328 134934061111104 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0417 04:06:44.632730 134934061111104 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10. I0417 04:06:45.349891 134934061111104 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 04:06:45.362672 134934061111104 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:06:45.362744 134934061111104 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 04:06:45.631333 134934061111104 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269486s I0417 04:06:45.631862 134812018919168 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 04:06:45.632404 134934061111104 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.622 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.274385929107666 s) (per-host) I0417 04:06:45.632482 134934061111104 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.274485s (batch_requests_ready=0.002074s, total_serialization_initiated=0.271414s, others=0.000997s) I0417 04:06:45.632564 134934061111104 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.278637s (all_items=0.000022s, per_item={'items': '0.00002170'}, temp_paths=0.278615) I0417 04:06:45.633326 134934061111104 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.00 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10. I0417 04:06:45.633623 134813067482880 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:26:45.633590 I0417 04:06:45.643617 134934061111104 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:06:45.643878 134808808367872 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:06:45.643999 134934061111104 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776398804.6293252, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776398804.6313672, 'checkpointer_blocking_duration_secs': 1.0024032592773438, 'get_old_steps_start_time': 1776398805.6337929, 'get_old_steps_duration_secs': 3.9577484130859375e-05, 'checkpoint_manager_blocking_start_time': 1776398804.6275938, 'checkpoint_manager_blocking_duration_secs': 1.016366720199585} I0417 04:06:45.644128 134934061111104 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 04:06:45.644195 134934061111104 max_utils.py:750] Memstats: After params initialized: I0417 04:06:45.644267 134934061111104 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 04:06:45.644321 134934061111104 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 04:06:45.644368 134934061111104 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 04:06:45.644411 134934061111104 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 04:06:45.980056 134934061111104 metric_logger.py:185] completed step: 10, seconds: 11.546, TFLOP/s/device: 1.177, Tokens/s/device: 177.373, total_weights: 65536, loss: 6.833 I0417 04:06:46.125393 134934061111104 metric_logger.py:185] completed step: 11, seconds: 1.505, TFLOP/s/device: 9.027, Tokens/s/device: 1360.632, total_weights: 65536, loss: 6.527 I0417 04:06:46.272711 134934061111104 metric_logger.py:185] completed step: 12, seconds: 0.010, TFLOP/s/device: 1375.494, Tokens/s/device: 207329.419, total_weights: 65536, loss: 6.265 I0417 04:06:46.420213 134934061111104 metric_logger.py:185] completed step: 13, seconds: 0.146, TFLOP/s/device: 93.312, Tokens/s/device: 14064.968, total_weights: 65536, loss: 6.045 I0417 04:06:46.429409 134934061111104 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 04:06:46.714644 134934061111104 checkpointing.py:798] Waited 0.2850801944732666 seconds for step 15 to finish before starting checkpointing. I0417 04:06:46.717704 134934061111104 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:06:48.296530 134809342957312 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 04:06:48.426561 134808799975168 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776398808020280936, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776398808020280936, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 04:06:51.188340 134809326171904 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0417 04:07:31.012822 134808816760576 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.602205s (commit=40.401388s, array_metadata_write=1.200817s) I0417 04:07:31.014382 134813067482880 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.598 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.656325578689575 s) (per-host) I0417 04:07:31.014553 134813067482880 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.380812s. I0417 04:07:32.470033 134808799975168 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776398808020280936, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 04:07:32.517208 134813067482880 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000396s. I0417 04:07:32.715561 134813067482880 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:07:33.816561 134813067482880 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.918305s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 04:07:33.818865 134813067482880 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 04:07:34.372342 134813067482880 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 04:07:35.768712 134813067482880 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10`. I0417 04:07:35.769877 134813067482880 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.14 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 04:07:35.771188 134813067482880 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.137445s. I0417 04:07:35.771368 134808808367872 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:07:35.771507 134808808367872 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:07:35.771635 134808808367872 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:07:35.773400 134808808367872 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 04:07:35.773725 134934061111104 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0417 04:07:35.773931 134934061111104 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.056226 seconds. If this number is high, consider checkpointing less frequently. I0417 04:07:35.775663 134934061111104 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0417 04:07:35.778052 134934061111104 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15. I0417 04:07:36.512149 134934061111104 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:07:36.512285 134934061111104 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 04:07:36.540174 134934061111104 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029086s I0417 04:07:36.541015 134934061111104 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.675 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03792452812194824 s) (per-host) I0417 04:07:36.541118 134934061111104 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038052s (batch_requests_ready=0.004506s, total_serialization_initiated=0.032772s, others=0.000774s) I0417 04:07:36.541241 134934061111104 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043035s (all_items=0.000024s, per_item={'items': '0.00002360'}, temp_paths=0.043011) I0417 04:07:36.542167 134934061111104 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15. I0417 04:07:36.542561 134808808367872 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:27:36.542520 I0417 04:07:36.545061 134934061111104 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:07:36.545428 134809334564608 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:07:36.545657 134934061111104 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776398806.7176597, 'wait_for_prev_duration_secs': 49.0562264919281, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776398855.7757304, 'checkpointer_blocking_duration_secs': 0.7670097351074219, 'get_old_steps_start_time': 1776398856.542766, 'get_old_steps_duration_secs': 4.4345855712890625e-05, 'checkpoint_manager_blocking_start_time': 1776398806.7154186, 'checkpoint_manager_blocking_duration_secs': 49.83018517494202} I0417 04:07:36.545807 134934061111104 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 04:07:36.731186 134813067482880 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 04:07:38.468204 134809326171904 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 04:07:43.122648 134809351350016 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0417 04:07:57.904631 134934061111104 metric_logger.py:185] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.143, Tokens/s/device: 13888.795, total_weights: 65536, loss: 5.864 I0417 04:07:57.913651 134934061111104 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.909, Tokens/s/device: 13853.470, total_weights: 65536, loss: 5.719 I0417 04:07:58.059147 134934061111104 metric_logger.py:185] completed step: 16, seconds: 71.483, TFLOP/s/device: 0.190, Tokens/s/device: 28.650, total_weights: 65536, loss: 5.605 I0417 04:07:58.206434 134934061111104 metric_logger.py:185] completed step: 17, seconds: 0.007, TFLOP/s/device: 1926.706, Tokens/s/device: 290414.067, total_weights: 65536, loss: 5.517 I0417 04:07:58.354193 134934061111104 metric_logger.py:185] completed step: 18, seconds: 0.147, TFLOP/s/device: 92.420, Tokens/s/device: 13930.551, total_weights: 65536, loss: 5.448 I0417 04:07:58.501218 134934061111104 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 04:07:58.501982 134934061111104 checkpointing.py:798] Waited 0.0007843971252441406 seconds for step 19 to finish before starting checkpointing. I0417 04:07:58.503979 134934061111104 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:08:22.888397 134809342957312 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.369569s (commit=40.308569s, array_metadata_write=2.061000s) I0417 04:08:22.889400 134808808367872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.053 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.38629508018494 s) (per-host) I0417 04:08:22.889523 134808808367872 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.346822s. I0417 04:08:25.233713 134808808367872 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:08:25.827548 134808808367872 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000228s. I0417 04:08:26.158224 134808808367872 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.161482s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 04:08:26.159739 134808808367872 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 04:08:26.715901 134808808367872 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 04:08:28.532612 134808808367872 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15`. I0417 04:08:28.533475 134808808367872 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.76 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 04:08:28.535545 134808808367872 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.992841s. I0417 04:08:28.535749 134809334564608 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:08:28.535878 134809334564608 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:08:28.535953 134809334564608 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:08:28.537663 134809334564608 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 04:08:28.537865 134934061111104 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0417 04:08:28.537997 134934061111104 checkpoint_manager.py:1452] Waiting for previous save to complete took 30.034022 seconds. If this number is high, consider checkpointing less frequently. I0417 04:08:28.539848 134934061111104 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0417 04:08:28.541861 134934061111104 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19. I0417 04:08:29.262536 134934061111104 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0417 04:08:29.262630 134934061111104 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 04:08:29.288353 134934061111104 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026550s I0417 04:08:29.289039 134934061111104 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 49.684 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031047821044921875 s) (per-host) I0417 04:08:29.289104 134934061111104 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031129s (batch_requests_ready=0.001900s, total_serialization_initiated=0.028593s, others=0.000636s) I0417 04:08:29.289177 134934061111104 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035112s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.035096) I0417 04:08:29.289877 134934061111104 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19. I0417 04:08:29.290173 134809334564608 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:28:29.290140 I0417 04:08:29.292114 134934061111104 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:08:29.292347 134809342957312 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:08:29.292450 134934061111104 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776398878.503945, 'wait_for_prev_duration_secs': 30.034022331237793, 'time_between_consecutive_saves_sec': 22.730477809906006, 'checkpointer_blocking_start_time': 1776398908.5398915, 'checkpointer_blocking_duration_secs': 0.7504186630249023, 'get_old_steps_start_time': 1776398909.2903287, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776398878.5022233, 'checkpoint_manager_blocking_duration_secs': 30.79019546508789} I0417 04:08:29.292609 134934061111104 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 04:08:29.292653 134934061111104 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:08:29.505907 134808808367872 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 04:08:31.664169 134811993741056 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 04:08:34.895996 134812018919168 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0417 04:09:14.519793 134809351350016 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.366759s (commit=40.175951s, array_metadata_write=1.190808s) I0417 04:09:14.520932 134809334564608 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.898 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.26288938522339 s) (per-host) I0417 04:09:14.521055 134809334564608 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.230767s. I0417 04:09:16.183689 134809334564608 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:09:16.363886 134809334564608 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000265s. I0417 04:09:17.241304 134809334564608 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.899760s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 04:09:17.242979 134809334564608 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 04:09:18.204447 134809334564608 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 04:09:20.050081 134809334564608 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19`. I0417 04:09:20.050976 134809334564608 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.51 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 04:09:20.052446 134809334564608 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.762156s. I0417 04:09:20.052577 134809342957312 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:09:20.052640 134809342957312 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:09:20.052690 134809342957312 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:09:20.054483 134809342957312 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 04:09:20.054636 134934061111104 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0417 04:09:20.054779 134934061111104 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:09:20.055411 134934061111104 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.532, Tokens/s/device: 13947.438, total_weights: 65536, loss: 5.391 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 04:09:32 UTC 2026 EXIT_CODE=0