XPK Start: Sun Apr 19 01:01:04 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 01:01:28.229715: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0419 01:01:28.411066 134859853043520 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 01:01:37.453689 134859853043520 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 01:01:37.458215 134859853043520 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-u8wpg-slice-job-0-0.mt-15-ocdbt-false-rs-u8wpg:8482 I0419 01:01:39.541769 134859853043520 max_utils.py:284] Jax distributed system initialized! I0419 01:01:44.656789 134859853043520 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 01:01:44.656906 134859853043520 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 01:01:44.656946 134859853043520 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0419 01:01:44.656981 134859853043520 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 01:01:45.711972 134859853043520 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 01:01:45.712560 134859853043520 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 01:01:45.712741 134859853043520 checkpointing.py:688] Setting up checkpoint logger... I0419 01:01:45.712794 134859853043520 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0419 01:01:45.712861 134859853043520 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 01:01:45.713192 134859853043520 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7aa6e631f830>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 01:01:46.949254 134859853043520 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 01:01:46.949486 134859853043520 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 0x7aa5be1972f0>}, handler_registry=None I0419 01:01:46.949717 134859853043520 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa5be1972f0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0419 01:01:46.949767 134859853043520 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa5be3071a0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0419 01:01:46.949803 134859853043520 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 0x7aa5be1972f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa5be1972f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa5be3071a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa5be3071a0>}). I0419 01:01:46.950154 134859853043520 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 01:01:46.950230 134859853043520 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a91d428ac00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 01:01:50.185652 134738425673472 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776560374622055552, 'commit_timestamp_nsecs': 1776560388332297082, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/_CHECKPOINT_METADATA I0419 01:01:50.432512 134738425673472 checkpoint_manager.py:1871] Missing metrics for step 9 I0419 01:01:50.511155 134859853043520 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints I0419 01:01:52.817079 134859853043520 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7aa5be195ca0> I0419 01:01:52.817251 134859853043520 checkpointing.py:302] Checkpoint manager created! I0419 01:01:53.744637 134859853043520 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:01:53.744739 134859853043520 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:01:54.129013 134859853043520 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:01:54.129108 134859853043520 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:01:54.145285 134859853043520 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:01:54.145346 134859853043520 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:01:54.168972 134859853043520 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:01:54.169038 134859853043520 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:01:54.185282 134859853043520 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:01:54.185343 134859853043520 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:01:54.201540 134859853043520 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:01:54.201600 134859853043520 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:01:54.226910 134859853043520 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 01:01:54.226984 134859853043520 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:01:54.247771 134859853043520 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 01:01:54.247853 134859853043520 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 01:01:54.469245 134859853043520 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 01:01:54.469354 134859853043520 checkpointing.py:582] restoring from this run's directory step 9 I0419 01:01:54.714424 134859853043520 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9. I0419 01:01:55.635967 134859853043520 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9. I0419 01:01:55.636398 134859853043520 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.92 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9. I0419 01:01:58.414827 1484 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 01:02:08.142789 134859853043520 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 143.931 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.979348421096802 s) (per-host) I0419 01:02:08.150989 134859853043520 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 139.064 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.38035798072815 s) (per-host) I0419 01:02:09.629031 134859853043520 event_tracking.py:138] [process=0] [sync] Finished load in 14.91 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9 I0419 01:02:09.629193 134859853043520 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776560514.7143276, 'checkpointer_duration_secs': 14.914825916290283, 'checkpoint_manager_start_time': 1776560514.4696083, 'checkpoint_manager_duration_secs': 15.159547090530396} fsdp: 32 I0419 01:02:10.045524 134859853043520 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0419 01:02:10.045647 134859853043520 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 01:02:10.045699 134859853043520 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 01:02:10.045757 134859853043520 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0419 01:02:10.045838 134859853043520 maxtext_utils.py:1827] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0419 01:02:10.045885 134859853043520 maxtext_utils.py:1827] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0419 01:02:10.045943 134859853043520 maxtext_utils.py:1827] 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) I0419 01:02:10.045997 134859853043520 maxtext_utils.py:1827] 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') I0419 01:02:10.046037 134859853043520 maxtext_utils.py:1827] 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) I0419 01:02:10.046075 134859853043520 maxtext_utils.py:1827] 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) I0419 01:02:10.046123 134859853043520 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0419 01:02:10.046174 134859853043520 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0419 01:02:10.540881 134859853043520 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 01:02:10.540977 134859853043520 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0419 01:02:10.556438 134859853043520 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 01:02:10.556502 134859853043520 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0419 01:02:21.485419 134859853043520 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0419 01:02:21.492879 134859853043520 metric_logger.py:301] number parameters: 1.104 billion I0419 01:02:33.266448 134859853043520 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0419 01:02:33.419386 134859853043520 checkpointing.py:798] Waited 0.15291905403137207 seconds for step 10 to finish before starting checkpointing. I0419 01:02:33.422007 134859853043520 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:02:33.424244 134859853043520 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0419 01:02:33.425719 134859853043520 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10. I0419 01:02:34.165750 134859853043520 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 01:02:34.178743 134859853043520 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0419 01:02:34.178823 134859853043520 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0419 01:02:34.449502 134859853043520 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.271630s I0419 01:02:34.450377 134859853043520 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.580 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2764420509338379 s) (per-host) I0419 01:02:34.450448 134859853043520 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.276538s (batch_requests_ready=0.002174s, total_serialization_initiated=0.273565s, others=0.000799s) I0419 01:02:34.450530 134859853043520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.280679s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.280661) I0419 01:02:34.451278 134859853043520 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10. I0419 01:02:34.451623 134735690962688 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:22:34.451586 I0419 01:02:34.466152 134859853043520 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:02:34.466399 134733577049856 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:02:34.466550 134859853043520 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776560553.4219878, 'wait_for_prev_duration_secs': 6.008148193359375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776560553.424283, 'checkpointer_blocking_duration_secs': 1.0274841785430908, 'get_old_steps_start_time': 1776560554.4517891, 'get_old_steps_duration_secs': 4.506111145019531e-05, 'checkpoint_manager_blocking_start_time': 1776560553.419928, 'checkpoint_manager_blocking_duration_secs': 1.046586275100708} I0419 01:02:34.466657 134859853043520 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0419 01:02:34.466707 134859853043520 max_utils.py:750] Memstats: After params initialized: I0419 01:02:34.466757 134859853043520 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0419 01:02:34.466792 134859853043520 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0419 01:02:34.466833 134859853043520 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0419 01:02:34.466860 134859853043520 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0419 01:02:34.800607 134859853043520 metric_logger.py:196] completed step: 10, seconds: 11.455, TFLOP/s/device: 1.186, Tokens/s/device: 178.791, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0419 01:02:34.838992 134734644680448 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10 I0419 01:02:34.945868 134859853043520 metric_logger.py:196] completed step: 11, seconds: 1.532, TFLOP/s/device: 8.867, Tokens/s/device: 1336.526, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0419 01:02:35.093416 134859853043520 metric_logger.py:196] completed step: 12, seconds: 0.010, TFLOP/s/device: 1311.372, Tokens/s/device: 197664.318, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0419 01:02:35.241028 134859853043520 metric_logger.py:196] completed step: 13, seconds: 0.146, TFLOP/s/device: 93.303, Tokens/s/device: 14063.712, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0419 01:02:35.250452 134859853043520 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0419 01:02:35.535527 134859853043520 checkpointing.py:798] Waited 0.2849717140197754 seconds for step 15 to finish before starting checkpointing. I0419 01:02:35.538544 134859853043520 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:02:37.036369 134734081046272 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/items I0419 01:02:37.202577 134733541889792 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776560556779685985, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776560556779685985, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA I0419 01:02:40.485938 134733593835264 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_0 I0419 01:02:47.068407 134733585442560 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 8.769415s (commit=7.689538s, array_metadata_write=1.079877s) I0419 01:02:47.069890 134735690962688 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 122.488 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 12.895951986312866 s) (per-host) I0419 01:02:47.069996 134735690962688 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 12.618259s. I0419 01:02:48.458538 134733541889792 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776560556779685985, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/_CHECKPOINT_METADATA W0419 01:02:48.514003 134735690962688 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/items, I0419 01:02:49.337203 134735690962688 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000412s. I0419 01:02:49.337446 134735690962688 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.471546s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/items I0419 01:02:49.339493 134735690962688 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10/items I0419 01:02:49.848188 134735690962688 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10 I0419 01:02:51.210212 134735690962688 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10`. I0419 01:02:51.211302 134735690962688 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 17.79 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/10 I0419 01:02:51.212772 134735690962688 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 16.761031s. I0419 01:02:51.212993 134733577049856 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:02:51.213109 134733577049856 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:02:51.213225 134733577049856 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:02:51.215108 134733577049856 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0419 01:02:51.215422 134859853043520 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0419 01:02:51.215595 134859853043520 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.677071 seconds. If this number is high, consider checkpointing less frequently. I0419 01:02:51.217475 134859853043520 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0419 01:02:51.220017 134859853043520 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15. I0419 01:02:51.940568 134859853043520 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0419 01:02:51.940703 134859853043520 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0419 01:02:51.970055 134859853043520 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.030552s I0419 01:02:51.971064 134859853043520 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 39.412 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.039139509201049805 s) (per-host) I0419 01:02:51.971176 134859853043520 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039279s (batch_requests_ready=0.003964s, total_serialization_initiated=0.034413s, others=0.000903s) I0419 01:02:51.971312 134859853043520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.044162s (all_items=0.000022s, per_item={'items': '0.00002170'}, temp_paths=0.044141) I0419 01:02:51.972278 134859853043520 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15. I0419 01:02:51.972705 134733577049856 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:22:51.972661 I0419 01:02:51.975136 134859853043520 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:02:51.975455 134733585442560 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:02:51.975658 134859853043520 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776560555.538497, 'wait_for_prev_duration_secs': 15.677070617675781, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776560571.2175615, 'checkpointer_blocking_duration_secs': 0.755342960357666, 'get_old_steps_start_time': 1776560571.9729316, 'get_old_steps_duration_secs': 4.6253204345703125e-05, 'checkpoint_manager_blocking_start_time': 1776560555.5361593, 'checkpoint_manager_blocking_duration_secs': 16.439445972442627} I0419 01:02:51.975804 134859853043520 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0419 01:02:52.169124 134735690962688 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15 I0419 01:02:53.913903 134732469737216 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15/items I0419 01:02:56.756335 134734072653568 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_0 I0419 01:03:04.628855 134733593835264 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 9.605243s (commit=8.034888s, array_metadata_write=1.570355s) I0419 01:03:04.629955 134733577049856 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 124.397 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 12.698014497756958 s) (per-host) I0419 01:03:04.630071 134733577049856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 12.657207s. W0419 01:03:06.528213 134733577049856 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15/items, I0419 01:03:07.366056 134733577049856 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000265s. I0419 01:03:07.366243 134733577049856 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.509569s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15/items I0419 01:03:07.367733 134733577049856 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15/items I0419 01:03:07.896045 134733577049856 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15 I0419 01:03:09.713789 134733577049856 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15`. I0419 01:03:09.714634 134733577049856 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.50 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/15 I0419 01:03:09.716070 134733577049856 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.743204s. I0419 01:03:09.716300 134733585442560 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:03:09.716422 134733585442560 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:03:09.716492 134733585442560 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:03:09.718318 134733585442560 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0419 01:03:12.765710 134859853043520 metric_logger.py:196] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.094, Tokens/s/device: 13881.452, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0419 01:03:12.775214 134859853043520 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.798, Tokens/s/device: 13836.809, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0419 01:03:12.920699 134859853043520 metric_logger.py:196] completed step: 16, seconds: 37.523, TFLOP/s/device: 0.362, Tokens/s/device: 54.580, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0419 01:03:13.068344 134859853043520 metric_logger.py:196] completed step: 17, seconds: 0.007, TFLOP/s/device: 1915.029, Tokens/s/device: 288653.982, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0419 01:03:13.215932 134859853043520 metric_logger.py:196] completed step: 18, seconds: 0.147, TFLOP/s/device: 92.568, Tokens/s/device: 13952.855, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0419 01:03:13.362677 134859853043520 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0419 01:03:13.363383 134859853043520 checkpointing.py:798] Waited 0.0007297992706298828 seconds for step 19 to finish before starting checkpointing. I0419 01:03:13.365425 134859853043520 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:03:13.367152 134859853043520 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0419 01:03:13.368677 134859853043520 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19. I0419 01:03:14.097752 134859853043520 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0419 01:03:14.097858 134859853043520 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0419 01:03:14.123553 134859853043520 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026515s I0419 01:03:14.124294 134859853043520 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 49.838 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030951976776123047 s) (per-host) I0419 01:03:14.124359 134859853043520 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031036s (batch_requests_ready=0.001836s, total_serialization_initiated=0.028503s, others=0.000698s) I0419 01:03:14.124428 134859853043520 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034963s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.034946) I0419 01:03:14.125146 134859853043520 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19. I0419 01:03:14.125421 134733585442560 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 01:23:14.125399 I0419 01:03:14.128139 134859853043520 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:03:14.128358 134733593835264 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:03:14.128477 134859853043520 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776560593.3653905, 'wait_for_prev_duration_secs': 7.224082946777344e-05, 'time_between_consecutive_saves_sec': 3.647031784057617, 'checkpointer_blocking_start_time': 1776560593.367189, 'checkpointer_blocking_duration_secs': 0.7583272457122803, 'get_old_steps_start_time': 1776560594.125535, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776560593.3636231, 'checkpoint_manager_blocking_duration_secs': 0.7648212909698486} I0419 01:03:14.128579 134859853043520 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0419 01:03:14.128627 134859853043520 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:03:14.334051 134733577049856 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19 I0419 01:03:16.850450 134731983222528 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19/items I0419 01:03:20.527796 134734089438976 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_0 I0419 01:03:25.951478 134734072653568 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 8.026734s (commit=6.976307s, array_metadata_write=1.050427s) I0419 01:03:25.952601 134733585442560 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 133.196 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.859223365783691 s) (per-host) I0419 01:03:25.952719 134733585442560 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.827222s. W0419 01:03:27.356738 134733585442560 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19/items, I0419 01:03:28.170697 134733585442560 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000228s. I0419 01:03:28.170892 134733585442560 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.447331s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19/items I0419 01:03:28.172503 134733585442560 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19/items I0419 01:03:28.679748 134733585442560 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19 I0419 01:03:30.061127 134733585442560 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_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19`. I0419 01:03:30.061966 134733585442560 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 16.69 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/linen_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/19 I0419 01:03:30.063567 134733585442560 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 15.938067s. I0419 01:03:30.063713 134733593835264 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:03:30.063799 134733593835264 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:03:30.063881 134733593835264 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:03:30.065501 134733593835264 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0419 01:03:30.065677 134859853043520 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0419 01:03:30.065841 134859853043520 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:03:30.066844 134859853043520 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.962, Tokens/s/device: 13861.534, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 01:03:39 UTC 2026 EXIT_CODE=0