XPK Start: Mon Apr 20 18:58:56 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 18:59:20.556129: 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 18:59:20.734541 138098527274816 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 18:59:29.774359 138098527274816 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 18:59:29.776612 138098527274816 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-6befq-slice-job-0-0.mt-16-zarr3-false-rs-6befq:8482 I0420 18:59:31.409892 138098527274816 max_utils.py:284] Jax distributed system initialized! I0420 18:59:37.422695 138098527274816 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 18:59:37.422800 138098527274816 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 18:59:37.422839 138098527274816 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 18:59:37.422878 138098527274816 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 18:59:38.469223 138098527274816 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 18:59:38.582017 138098527274816 checkpointing.py:688] Setting up checkpoint logger... I0420 18:59:38.582213 138098527274816 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0420 18:59:38.582262 138098527274816 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 18:59:38.582456 138098527274816 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7d98d9f48890>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 18:59:39.826352 138098527274816 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 18:59:39.826586 138098527274816 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 0x7d97d2747a10>}, handler_registry=None I0420 18:59:39.826828 138098527274816 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d97d2747a10>` 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 18:59:39.826884 138098527274816 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d97cda323c0>` 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 18:59:39.826922 138098527274816 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 0x7d97d2747a10>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d97d2747a10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d97cda323c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d97cda323c0>}). I0420 18:59:39.827261 138098527274816 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 18:59:39.827331 138098527274816 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d85242e04a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 18:59:41.461604 137980538754816 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776711415694776946, 'commit_timestamp_nsecs': 1776711464057948737, '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_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0420 18:59:41.711629 137980538754816 checkpoint_manager.py:1871] Missing metrics for step 9 I0420 18:59:41.745482 138098527274816 checkpoint_manager.py:1812] Found 3 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_16_zarr3_false/checkpoints I0420 18:59:43.267663 138098527274816 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d97cdd0c8f0> I0420 18:59:43.267836 138098527274816 checkpointing.py:302] Checkpoint manager created! I0420 18:59:44.207620 138098527274816 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 18:59:44.207735 138098527274816 checkpointing.py:582] restoring from this run's directory step 9 I0420 18:59:44.459747 138098527274816 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/9. I0420 18:59:44.740171 138098527274816 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/9. I0420 18:59:44.740605 138098527274816 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.28 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/9. I0420 18:59:46.181131 1422 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 18:59:52.946729 138098527274816 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 233.335 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.772552728652954 s) (per-host) I0420 18:59:52.956382 138098527274816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 219.526 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.209197044372559 s) (per-host) I0420 18:59:54.184444 138098527274816 event_tracking.py:138] [process=0] [sync] Finished load in 9.72 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/9 I0420 18:59:54.184602 138098527274816 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776711584.4596486, 'checkpointer_duration_secs': 9.724920272827148, 'checkpoint_manager_start_time': 1776711584.2083433, 'checkpoint_manager_duration_secs': 9.976228475570679} fsdp: 32 I0420 18:59:54.260071 138098527274816 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 18:59:54.260194 138098527274816 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:59:54.260245 138098527274816 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:59:54.260288 138098527274816 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 18:59:54.260324 138098527274816 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:59:54.260358 138098527274816 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:59:54.260393 138098527274816 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:59:54.260427 138098527274816 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 18:59:54.260460 138098527274816 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:59:54.260491 138098527274816 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:59:54.260523 138098527274816 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 18:59:54.260555 138098527274816 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 18:59:54.665529 138098527274816 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:59:54.665618 138098527274816 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:59:54.670989 138098527274816 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:59:54.671047 138098527274816 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:59:54.687340 138098527274816 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:59:54.687401 138098527274816 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:59:54.702931 138098527274816 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:59:54.702988 138098527274816 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:59:54.726509 138098527274816 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 18:59:54.726574 138098527274816 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:59:54.742290 138098527274816 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 18:59:54.742351 138098527274816 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:59:54.757964 138098527274816 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 18:59:54.758024 138098527274816 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:59:54.787710 138098527274816 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 18:59:54.787780 138098527274816 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:59:54.807332 138098527274816 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 18:59:54.807394 138098527274816 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 19:00:05.951234 138098527274816 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 19:00:05.960452 138098527274816 metric_logger.py:301] number parameters: 1.104 billion I0420 19:00:17.510771 138098527274816 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 19:00:17.662961 138098527274816 checkpointing.py:798] Waited 0.15216970443725586 seconds for step 10 to finish before starting checkpointing. I0420 19:00:17.665410 138098527274816 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:00:17.667391 138098527274816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0420 19:00:17.668715 138098527274816 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_16_zarr3_false/checkpoints/10. I0420 19:00:17.991497 138098527274816 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 19:00:18.009261 138098527274816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 19:00:18.009379 138098527274816 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 19:00:18.256219 137979574085376 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_16_zarr3_false/checkpoints/10 I0420 19:00:18.283656 138098527274816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.275816s I0420 19:00:18.284491 138098527274816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.420 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2846200466156006 s) (per-host) I0420 19:00:18.284561 138098527274816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284707s (batch_requests_ready=0.002788s, total_serialization_initiated=0.281163s, others=0.000756s) I0420 19:00:18.284639 138098527274816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289048s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.289030) I0420 19:00:18.285378 138098527274816 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.62 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_16_zarr3_false/checkpoints/10. I0420 19:00:18.285671 137979582478080 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:20:18.285639 I0420 19:00:18.296971 138098527274816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:00:18.297243 137974884857600 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:00:18.297397 138098527274816 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776711617.6653917, 'wait_for_prev_duration_secs': 6.0558319091796875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776711617.667433, 'checkpointer_blocking_duration_secs': 0.6183786392211914, 'get_old_steps_start_time': 1776711618.2858407, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776711617.6633606, 'checkpoint_manager_blocking_duration_secs': 0.6340010166168213} I0420 19:00:18.297501 138098527274816 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 19:00:18.297568 138098527274816 max_utils.py:750] Memstats: After params initialized: I0420 19:00:18.297616 138098527274816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 19:00:18.297648 138098527274816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 19:00:18.297674 138098527274816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 19:00:18.297697 138098527274816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 19:00:18.662376 138098527274816 metric_logger.py:196] completed step: 10, seconds: 11.219, TFLOP/s/device: 1.211, Tokens/s/device: 182.548, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0420 19:00:18.803496 138098527274816 metric_logger.py:196] completed step: 11, seconds: 1.148, TFLOP/s/device: 11.835, Tokens/s/device: 1783.929, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0420 19:00:18.951416 138098527274816 metric_logger.py:196] completed step: 12, seconds: 0.014, TFLOP/s/device: 1002.370, Tokens/s/device: 151088.159, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0420 19:00:19.098990 138098527274816 metric_logger.py:196] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.621, Tokens/s/device: 14413.096, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0420 19:00:19.111628 138098527274816 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0420 19:00:19.393607 138098527274816 checkpointing.py:798] Waited 0.2819359302520752 seconds for step 15 to finish before starting checkpointing. I0420 19:00:19.396341 138098527274816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:00:20.239697 137974876464896 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776711619833116339, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776711619833116339, "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_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0420 19:00:20.527597 137975783151360 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_16_zarr3_false/checkpoints/10/items I0420 19:00:23.756516 137975766365952 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/10/items/array_metadatas/process_0 I0420 19:01:03.756730 137975229249280 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.849737s (commit=40.517225s, array_metadata_write=1.332512s) I0420 19:01:03.758216 137979582478080 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.520 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.75833511352539 s) (per-host) I0420 19:01:03.758322 137979582478080 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.472540s. I0420 19:01:05.280477 137974876464896 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776711619833116339, '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_16_zarr3_false/checkpoints/10/_CHECKPOINT_METADATA I0420 19:01:05.292094 137979582478080 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000719s. I0420 19:01:06.648520 137979582478080 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.969767s. use_zarr3=False, 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_16_zarr3_false/checkpoints/10/items I0420 19:01:06.651711 137979582478080 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_16_zarr3_false/checkpoints/10/items I0420 19:01:07.286508 137979582478080 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_16_zarr3_false/checkpoints/10 I0420 19:01:08.824461 137979582478080 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_16_zarr3_false/checkpoints/10`. I0420 19:01:08.825580 137979582478080 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.16 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/10 I0420 19:01:08.827144 137979582478080 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.541358s. I0420 19:01:08.827387 137974884857600 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:01:08.827498 137974884857600 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:01:08.827604 137974884857600 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:01:08.829171 137974884857600 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 19:01:08.829504 138098527274816 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 19:01:08.829715 138098527274816 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.433374 seconds. If this number is high, consider checkpointing less frequently. I0420 19:01:08.831705 138098527274816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0420 19:01:08.834250 138098527274816 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_16_zarr3_false/checkpoints/15. I0420 19:01:09.149950 138098527274816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 19:01:09.150103 138098527274816 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 19:01:09.181553 138098527274816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034530s I0420 19:01:09.182526 138098527274816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 28.603 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05393075942993164 s) (per-host) I0420 19:01:09.182632 138098527274816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.054061s (batch_requests_ready=0.006729s, total_serialization_initiated=0.046461s, others=0.000871s) I0420 19:01:09.182763 138098527274816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.059057s (all_items=0.000023s, per_item={'items': '0.00002265'}, temp_paths=0.059035) I0420 19:01:09.183836 138098527274816 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.35 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_16_zarr3_false/checkpoints/15. I0420 19:01:09.184327 137974884857600 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:21:09.184285 I0420 19:01:09.186851 138098527274816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:01:09.187220 137975774758656 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:01:09.187448 138098527274816 standard_logger.py:34] {'step': 15, '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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776711619.3962953, 'wait_for_prev_duration_secs': 49.43337416648865, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776711668.831771, 'checkpointer_blocking_duration_secs': 0.3526918888092041, 'get_old_steps_start_time': 1776711669.184494, 'get_old_steps_duration_secs': 4.410743713378906e-05, 'checkpoint_manager_blocking_start_time': 1776711619.3940988, 'checkpoint_manager_blocking_duration_secs': 49.793296337127686} I0420 19:01:09.187603 138098527274816 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0420 19:01:09.378062 137979582478080 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_16_zarr3_false/checkpoints/15 I0420 19:01:11.513142 137975766365952 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_16_zarr3_false/checkpoints/15/items I0420 19:01:14.515111 137978508736256 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/15/items/array_metadatas/process_0 I0420 19:01:30.601129 138098527274816 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.172, Tokens/s/device: 13893.129, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0420 19:01:30.612357 138098527274816 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.699, Tokens/s/device: 13821.868, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0420 19:01:30.756593 138098527274816 metric_logger.py:196] completed step: 16, seconds: 71.500, TFLOP/s/device: 0.190, Tokens/s/device: 28.643, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0420 19:01:30.903757 138098527274816 metric_logger.py:196] completed step: 17, seconds: 0.009, TFLOP/s/device: 1497.369, Tokens/s/device: 225699.802, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0420 19:01:31.051951 138098527274816 metric_logger.py:196] completed step: 18, seconds: 0.145, TFLOP/s/device: 93.453, Tokens/s/device: 14086.251, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0420 19:01:31.199700 138098527274816 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 19:01:31.200459 138098527274816 checkpointing.py:798] Waited 0.0007708072662353516 seconds for step 19 to finish before starting checkpointing. I0420 19:01:31.202499 138098527274816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:01:54.588204 137975783151360 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.900873s (commit=40.630708s, array_metadata_write=1.270165s) I0420 19:01:54.589265 137974884857600 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.747 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.46065282821655 s) (per-host) I0420 19:01:54.589398 137974884857600 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.404233s. I0420 19:01:56.377845 137974884857600 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000379s. I0420 19:01:57.457667 137974884857600 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.935524s. use_zarr3=False, 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_16_zarr3_false/checkpoints/15/items I0420 19:01:57.459425 137974884857600 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_16_zarr3_false/checkpoints/15/items I0420 19:01:58.002831 137974884857600 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_16_zarr3_false/checkpoints/15 I0420 19:01:59.419902 137974884857600 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_16_zarr3_false/checkpoints/15`. I0420 19:01:59.420755 137974884857600 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.59 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/15 I0420 19:01:59.422312 137974884857600 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.237144s. I0420 19:01:59.422526 137975774758656 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:01:59.422657 137975774758656 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:01:59.422725 137975774758656 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:01:59.427765 137975774758656 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0420 19:01:59.427931 138098527274816 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. W0420 19:01:59.428056 138098527274816 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.225566 seconds. If this number is high, consider checkpointing less frequently. I0420 19:01:59.429594 138098527274816 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0420 19:01:59.431551 138098527274816 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_16_zarr3_false/checkpoints/19. I0420 19:01:59.741068 138098527274816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 19:01:59.741174 138098527274816 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 19:01:59.768552 138098527274816 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028935s I0420 19:01:59.769328 138098527274816 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.046 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037581443786621094 s) (per-host) I0420 19:01:59.769388 138098527274816 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037658s (batch_requests_ready=0.002769s, total_serialization_initiated=0.034202s, others=0.000688s) I0420 19:01:59.769467 138098527274816 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041837s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.041820) I0420 19:01:59.770220 138098527274816 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.34 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_16_zarr3_false/checkpoints/19. I0420 19:01:59.770521 137975774758656 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 19:21:59.770483 I0420 19:01:59.772562 138098527274816 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:01:59.772830 137978995250944 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:01:59.772965 138098527274816 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_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776711691.202462, 'wait_for_prev_duration_secs': 28.225566148757935, 'time_between_consecutive_saves_sec': 22.373222589492798, 'checkpointer_blocking_start_time': 1776711719.4296327, 'checkpointer_blocking_duration_secs': 0.34106945991516113, 'get_old_steps_start_time': 1776711719.7707229, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776711691.200686, 'checkpoint_manager_blocking_duration_secs': 28.57224750518799} I0420 19:01:59.773134 138098527274816 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 19:01:59.773177 138098527274816 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:01:59.979523 137974884857600 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_16_zarr3_false/checkpoints/19 I0420 19:02:01.762197 137975783151360 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_16_zarr3_false/checkpoints/19/items I0420 19:02:04.775581 137979574085376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/19/items/array_metadatas/process_0 I0420 19:02:48.213250 137979003643648 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 45.289597s (commit=44.069905s, array_metadata_write=1.219692s) I0420 19:02:48.215573 137975774758656 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 32.580 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 48.48378944396973 s) (per-host) I0420 19:02:48.215684 137975774758656 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 48.445014s. I0420 19:02:49.754028 137975774758656 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000401s. I0420 19:02:50.596385 137975774758656 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.489003s. use_zarr3=False, 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_16_zarr3_false/checkpoints/19/items I0420 19:02:50.598158 137975774758656 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_16_zarr3_false/checkpoints/19/items I0420 19:02:51.157196 137975774758656 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_16_zarr3_false/checkpoints/19 I0420 19:02:52.632977 137975774758656 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_16_zarr3_false/checkpoints/19`. I0420 19:02:52.633829 137975774758656 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 53.20 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_16_zarr3_false/checkpoints/19 I0420 19:02:52.635704 137975774758656 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.865021s. I0420 19:02:52.635926 137978995250944 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:02:52.636000 137978995250944 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:02:52.636048 137978995250944 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:02:52.638502 137978995250944 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 19:02:52.638649 138098527274816 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 19:02:52.638826 138098527274816 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:02:52.639783 138098527274816 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.416, Tokens/s/device: 13929.888, total_weights: 65536, loss: 5.385, lm_loss: 5.385, perplexity: 218.045 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 19:03:02 UTC 2026 EXIT_CODE=0