XPK Start: Fri Apr 17 06:05:27 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 06:05:51.362748: 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 06:05:51.895971 139103455283008 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 06:06:00.937340 139103455283008 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 06:06:00.939586 139103455283008 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-5vkqu-slice-job-0-0.mt-15-ocdbt-true-rs-5vkqu:8482 I0417 06:06:02.016079 139103455283008 max_utils.py:284] Jax distributed system initialized! I0417 06:06:08.237181 139103455283008 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 06:06:08.237283 139103455283008 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 06:06:08.237332 139103455283008 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 06:06:08.237369 139103455283008 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 06:06:08.904055 139103455283008 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 06:06:09.017966 139103455283008 checkpointing.py:688] Setting up checkpoint logger... I0417 06:06:09.018077 139103455283008 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 06:06:09.018121 139103455283008 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 06:06:09.018314 139103455283008 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 0x7e82ee119d00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 06:06:10.346135 139103455283008 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 06:06:10.346378 139103455283008 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 0x7e6e687e69f0>}, handler_registry=None I0417 06:06:10.346657 139103455283008 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6e687e69f0>` 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 06:06:10.346723 139103455283008 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6e885b4050>` 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 06:06:10.346775 139103455283008 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 0x7e6e687e69f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6e687e69f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6e885b4050>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e6e885b4050>}). I0417 06:06:10.347150 139103455283008 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 06:06:10.347226 139103455283008 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e6ea8281ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 06:06:12.669375 138981635249920 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776405802579295779, 'commit_timestamp_nsecs': 1776405849453734637, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9/_CHECKPOINT_METADATA I0417 06:06:12.997166 138981635249920 checkpoint_manager.py:1871] Missing metrics for step 9 I0417 06:06:13.419115 139103455283008 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints I0417 06:06:14.284522 139103455283008 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e6e885acbc0> I0417 06:06:14.284694 139103455283008 checkpointing.py:302] Checkpoint manager created! I0417 06:06:15.220862 139103455283008 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 06:06:15.220978 139103455283008 checkpointing.py:582] restoring from this run's directory step 9 I0417 06:06:15.473235 139103455283008 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 06:06:15.756696 139103455283008 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 06:06:15.757138 139103455283008 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9. I0417 06:06:17.148721 1457 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 06:06:24.093027 139103455283008 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 227.330 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.951452732086182 s) (per-host) I0417 06:06:24.102830 139103455283008 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 215.216 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.353557348251343 s) (per-host) I0417 06:06:25.891354 139103455283008 event_tracking.py:138] [process=0] [sync] Finished load in 10.42 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/9 I0417 06:06:25.891510 139103455283008 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776405975.4731324, 'checkpointer_duration_secs': 10.418347358703613, 'checkpoint_manager_start_time': 1776405975.221611, 'checkpoint_manager_duration_secs': 10.669871091842651} fsdp: 32 I0417 06:06:25.967735 139103455283008 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 06:06:25.967855 139103455283008 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:06:25.967909 139103455283008 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:06:25.967951 139103455283008 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 06:06:25.967988 139103455283008 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:06:25.968023 139103455283008 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:06:25.968060 139103455283008 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:06:25.968104 139103455283008 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 06:06:25.968145 139103455283008 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:06:25.968178 139103455283008 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:06:25.968211 139103455283008 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 06:06:25.968244 139103455283008 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 06:06:26.370056 139103455283008 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:06:26.370149 139103455283008 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:06:26.375622 139103455283008 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:06:26.375677 139103455283008 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:06:26.391971 139103455283008 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:06:26.392026 139103455283008 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:06:26.407357 139103455283008 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:06:26.407415 139103455283008 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:06:26.431062 139103455283008 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 06:06:26.431137 139103455283008 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:06:26.446690 139103455283008 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 06:06:26.446747 139103455283008 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:06:26.462167 139103455283008 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 06:06:26.462223 139103455283008 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:06:26.491534 139103455283008 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 06:06:26.491603 139103455283008 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:06:26.511393 139103455283008 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 06:06:26.511462 139103455283008 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 06:06:37.579632 139103455283008 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 06:06:37.589037 139103455283008 metric_logger.py:289] number parameters: 1.104 billion I0417 06:06:49.042668 139103455283008 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 06:06:49.195185 139103455283008 checkpointing.py:798] Waited 0.15249943733215332 seconds for step 10 to finish before starting checkpointing. I0417 06:06:49.197531 139103455283008 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:06:49.199328 139103455283008 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0417 06:06:49.200892 139103455283008 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10. I0417 06:06:49.921489 139103455283008 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 06:06:49.938500 139103455283008 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:06:49.938619 139103455283008 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 06:06:50.185568 138980460828416 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 06:06:50.209828 139103455283008 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272747s I0417 06:06:50.210704 139103455283008 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.478 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28157639503479004 s) (per-host) I0417 06:06:50.210790 139103455283008 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.281685s (batch_requests_ready=0.002889s, total_serialization_initiated=0.278002s, others=0.000794s) I0417 06:06:50.210951 139103455283008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285693s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.285675) I0417 06:06:50.211694 139103455283008 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10. I0417 06:06:50.211970 138981031270144 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:26:50.211947 I0417 06:06:50.227025 139103455283008 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:06:50.227335 138976711124736 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:06:50.227484 139103455283008 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_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': 1776406009.1975124, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406009.1993654, 'checkpointer_blocking_duration_secs': 1.0127053260803223, 'get_old_steps_start_time': 1776406010.212093, 'get_old_steps_duration_secs': 3.266334533691406e-05, 'checkpoint_manager_blocking_start_time': 1776406009.1955466, 'checkpoint_manager_blocking_duration_secs': 1.0319035053253174} I0417 06:06:50.227586 139103455283008 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 06:06:50.227656 139103455283008 max_utils.py:750] Memstats: After params initialized: I0417 06:06:50.227706 139103455283008 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 06:06:50.227738 139103455283008 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 06:06:50.227764 139103455283008 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 06:06:50.227787 139103455283008 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 06:06:50.583622 139103455283008 metric_logger.py:185] completed step: 10, seconds: 11.148, TFLOP/s/device: 1.219, Tokens/s/device: 183.706, total_weights: 65536, loss: 6.827 I0417 06:06:50.724572 139103455283008 metric_logger.py:185] completed step: 11, seconds: 1.538, TFLOP/s/device: 8.836, Tokens/s/device: 1331.927, total_weights: 65536, loss: 6.522 I0417 06:06:50.872474 139103455283008 metric_logger.py:185] completed step: 12, seconds: 0.014, TFLOP/s/device: 993.865, Tokens/s/device: 149806.159, total_weights: 65536, loss: 6.260 I0417 06:06:51.020242 139103455283008 metric_logger.py:185] completed step: 13, seconds: 0.142, TFLOP/s/device: 95.762, Tokens/s/device: 14434.225, total_weights: 65536, loss: 6.039 I0417 06:06:51.033167 139103455283008 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 06:06:51.315377 139103455283008 checkpointing.py:798] Waited 0.28217530250549316 seconds for step 15 to finish before starting checkpointing. I0417 06:06:51.318247 139103455283008 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:06:52.436872 138976753088256 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 06:06:52.610869 138976174954240 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776406012183741623, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776406012183741623, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 06:06:55.677650 138976727910144 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0417 06:07:34.890656 138976719517440 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.321779s (commit=40.118411s, array_metadata_write=1.203368s) I0417 06:07:34.892131 138981031270144 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.131 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.9629693031311 s) (per-host) I0417 06:07:34.892302 138981031270144 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.680260s. I0417 06:07:36.394578 138976174954240 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776406012183741623, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 06:07:36.427125 138981031270144 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000780s. I0417 06:07:36.585235 138981031270144 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 06:07:37.681204 138981031270144 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.919004s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 06:07:37.683497 138981031270144 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10/items I0417 06:07:38.483126 138981031270144 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 06:07:39.906242 138981031270144 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10`. I0417 06:07:39.907339 138981031270144 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.71 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/10 I0417 06:07:39.909003 138981031270144 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.696958s. I0417 06:07:39.909245 138976711124736 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:07:39.909376 138976711124736 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:07:39.909503 138976711124736 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:07:39.911594 138976711124736 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 06:07:39.911943 139103455283008 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 06:07:39.912053 139103455283008 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.593814 seconds. If this number is high, consider checkpointing less frequently. I0417 06:07:39.913736 139103455283008 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0417 06:07:39.916117 139103455283008 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15. I0417 06:07:40.643239 139103455283008 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:07:40.643376 139103455283008 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 06:07:40.674139 139103455283008 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.033915s I0417 06:07:40.675063 139103455283008 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 28.939 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05330467224121094 s) (per-host) I0417 06:07:40.675165 139103455283008 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.053430s (batch_requests_ready=0.006659s, total_serialization_initiated=0.045914s, others=0.000857s) I0417 06:07:40.675284 139103455283008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.058405s (all_items=0.000022s, per_item={'items': '0.00002241'}, temp_paths=0.058383) I0417 06:07:40.676265 139103455283008 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15. I0417 06:07:40.676707 138976711124736 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:27:40.676665 I0417 06:07:40.679184 139103455283008 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:07:40.679536 138976727910144 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:07:40.679795 139103455283008 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_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': 1776406011.3181999, 'wait_for_prev_duration_secs': 48.59381437301636, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406059.9138017, 'checkpointer_blocking_duration_secs': 0.7631072998046875, 'get_old_steps_start_time': 1776406060.6769452, 'get_old_steps_duration_secs': 4.4345855712890625e-05, 'checkpoint_manager_blocking_start_time': 1776406011.315895, 'checkpoint_manager_blocking_duration_secs': 49.36383056640625} I0417 06:07:40.680015 139103455283008 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 06:07:40.867784 138981031270144 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 06:07:43.093041 138980460828416 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 06:07:45.979275 138976753088256 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0417 06:08:02.179508 139103455283008 metric_logger.py:185] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.083, Tokens/s/device: 13879.758, total_weights: 65536, loss: 5.858 I0417 06:08:02.190333 139103455283008 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.752, Tokens/s/device: 13829.895, total_weights: 65536, loss: 5.713 I0417 06:08:02.334289 139103455283008 metric_logger.py:185] completed step: 16, seconds: 71.157, TFLOP/s/device: 0.191, Tokens/s/device: 28.781, total_weights: 65536, loss: 5.599 I0417 06:08:02.481702 139103455283008 metric_logger.py:185] completed step: 17, seconds: 0.008, TFLOP/s/device: 1603.958, Tokens/s/device: 241766.025, total_weights: 65536, loss: 5.511 I0417 06:08:02.629443 139103455283008 metric_logger.py:185] completed step: 18, seconds: 0.146, TFLOP/s/device: 93.301, Tokens/s/device: 14063.326, total_weights: 65536, loss: 5.442 I0417 06:08:02.777790 139103455283008 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 06:08:02.778548 139103455283008 checkpointing.py:798] Waited 0.0007696151733398438 seconds for step 19 to finish before starting checkpointing. I0417 06:08:02.780469 139103455283008 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:08:25.927395 138976744695552 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.731137s (commit=40.070365s, array_metadata_write=1.660772s) I0417 06:08:25.928596 138976711124736 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.864 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.30681848526001 s) (per-host) I0417 06:08:25.928726 138976711124736 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.251859s. I0417 06:08:27.385221 138976711124736 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000403s. I0417 06:08:27.628343 138976711124736 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 06:08:28.721753 138976711124736 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.926048s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 06:08:28.723421 138976711124736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15/items I0417 06:08:29.267723 138976711124736 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 06:08:30.668008 138976711124736 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15`. I0417 06:08:30.668818 138976711124736 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.75 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/15 I0417 06:08:30.670377 138976711124736 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.993508s. I0417 06:08:30.670596 138976727910144 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:08:30.670729 138976727910144 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:08:30.670808 138976727910144 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:08:30.673373 138976727910144 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 06:08:30.673546 139103455283008 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 06:08:30.673666 139103455283008 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.893205 seconds. If this number is high, consider checkpointing less frequently. I0417 06:08:30.675225 139103455283008 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0417 06:08:30.677283 139103455283008 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19. I0417 06:08:30.989517 139103455283008 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:08:30.989614 139103455283008 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 06:08:31.016695 139103455283008 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028600s I0417 06:08:31.017532 139103455283008 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.308 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037343740463256836 s) (per-host) I0417 06:08:31.017611 139103455283008 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037444s (batch_requests_ready=0.002811s, total_serialization_initiated=0.033882s, others=0.000751s) I0417 06:08:31.017699 139103455283008 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041612s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.041594) I0417 06:08:31.018389 139103455283008 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19. I0417 06:08:31.018706 138977238828800 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:28:31.018666 I0417 06:08:31.020649 139103455283008 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:08:31.020956 138976727910144 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:08:31.021125 139103455283008 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_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': 1776406082.7804344, 'wait_for_prev_duration_secs': 27.893205165863037, 'time_between_consecutive_saves_sec': 22.868712186813354, 'checkpointer_blocking_start_time': 1776406110.6752653, 'checkpointer_blocking_duration_secs': 0.3435990810394287, 'get_old_steps_start_time': 1776406111.018883, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776406082.7787838, 'checkpoint_manager_blocking_duration_secs': 28.24230933189392} I0417 06:08:31.021236 139103455283008 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 06:08:31.021279 139103455283008 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:08:31.222495 138976711124736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 06:08:33.405903 138976736302848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 06:08:36.799228 138980460828416 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0417 06:09:16.782683 138978346915584 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.772259s (commit=40.448577s, array_metadata_write=1.323682s) I0417 06:09:16.784042 138977238828800 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.486 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.80383610725403 s) (per-host) I0417 06:09:16.784154 138977238828800 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.765331s. I0417 06:09:18.443705 138977238828800 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 06:09:18.609611 138977238828800 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000416s. I0417 06:09:19.539041 138977238828800 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.911418s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 06:09:19.544377 138977238828800 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19/items I0417 06:09:20.085242 138977238828800 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 06:09:21.951663 138977238828800 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_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19`. I0417 06:09:21.952529 138977238828800 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.28 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_true/checkpoints/19 I0417 06:09:21.953716 138977238828800 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.934891s. I0417 06:09:21.953926 138976727910144 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:09:21.954038 138976727910144 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:09:21.954092 138976727910144 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:09:21.955949 138976727910144 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 06:09:21.956110 139103455283008 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 06:09:21.956273 139103455283008 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:09:21.956924 139103455283008 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.313, Tokens/s/device: 13914.461, total_weights: 65536, loss: 5.385 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 06:09:33 UTC 2026 EXIT_CODE=0