XPK Start: Fri Apr 17 02:01:17 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 02:01:40.966400: 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 02:01:41.143110 135477762000704 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:01:50.183925 135477762000704 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:01:50.186290 135477762000704 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-fdugr-slice-job-0-0.mt-15-ocdbt-true-rs-fdugr:8482 I0417 02:01:51.241567 135477762000704 max_utils.py:284] Jax distributed system initialized! I0417 02:01:57.342873 135477762000704 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:01:57.342974 135477762000704 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:01:57.343013 135477762000704 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 02:01:57.343048 135477762000704 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:01:58.002948 135477762000704 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:01:58.114682 135477762000704 checkpointing.py:688] Setting up checkpoint logger... I0417 02:01:58.114798 135477762000704 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 02:01:58.114864 135477762000704 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:01:58.115065 135477762000704 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 0x7b36d13a95e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:01:59.900567 135477762000704 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:01:59.900799 135477762000704 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 0x7b228460bfb0>}, handler_registry=None I0417 02:01:59.901053 135477762000704 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b228460bfb0>` 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 02:01:59.901107 135477762000704 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b22a445a180>` 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 02:01:59.901144 135477762000704 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 0x7b228460bfb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b228460bfb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b22a445a180>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b22a445a180>}). I0417 02:01:59.901469 135477762000704 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:01:59.901538 135477762000704 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b22bc0fe020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:02:01.817058 135356624135936 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776391056436179635, 'commit_timestamp_nsecs': 1776391098641711134, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0417 02:02:02.057965 135356624135936 checkpoint_manager.py:1871] Missing metrics for step 0 I0417 02:02:02.120805 135477762000704 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints I0417 02:02:03.706596 135477762000704 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b22bc12c7d0> I0417 02:02:03.706767 135477762000704 checkpointing.py:302] Checkpoint manager created! I0417 02:02:04.657240 135477762000704 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:02:04.657358 135477762000704 checkpointing.py:582] restoring from this run's directory step 9 I0417 02:02:04.910786 135477762000704 event_tracking.py:70] [process=0] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/9. I0417 02:02:05.154061 135477762000704 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/9. I0417 02:02:05.154479 135477762000704 event_tracking.py:125] [process=0] [sync] Finished blocking load in 0.24 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/9. I0417 02:02:06.969312 1445 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:02:13.381443 135477762000704 jax_array_handlers.py:843] [process=0] /jax/orbax/read/worker/io/requested throughput: 246.210 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.418418645858765 s) (per-host) I0417 02:02:13.390193 135477762000704 base_pytree_checkpoint_handler.py:130] [process=0] /jax/checkpoint/read/gbytes_per_sec: 232.134 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.817651033401489 s) (per-host) I0417 02:02:16.331973 135477762000704 event_tracking.py:138] [process=0] [sync] Finished load in 11.42 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/9 I0417 02:02:16.332127 135477762000704 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints', 'checkpointer_start_time': 1776391324.91069, 'checkpointer_duration_secs': 11.421403884887695, 'checkpoint_manager_start_time': 1776391324.6579707, 'checkpoint_manager_duration_secs': 11.674126148223877} fsdp: 32 I0417 02:02:16.406764 135477762000704 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:02:16.406889 135477762000704 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:02:16.406940 135477762000704 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:02:16.406985 135477762000704 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:02:16.407021 135477762000704 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:02:16.407055 135477762000704 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:02:16.407089 135477762000704 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:02:16.407124 135477762000704 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:02:16.407157 135477762000704 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:02:16.407188 135477762000704 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:02:16.407218 135477762000704 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:02:16.407250 135477762000704 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:02:16.805302 135477762000704 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:02:16.805392 135477762000704 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:02:16.810783 135477762000704 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:02:16.810855 135477762000704 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:02:16.826862 135477762000704 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:02:16.826921 135477762000704 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:02:16.842305 135477762000704 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:02:16.842365 135477762000704 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:02:16.865812 135477762000704 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 02:02:16.865890 135477762000704 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:02:16.881523 135477762000704 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 02:02:16.881584 135477762000704 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:02:16.897063 135477762000704 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 02:02:16.897121 135477762000704 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:02:16.926444 135477762000704 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:02:16.926517 135477762000704 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:02:16.945940 135477762000704 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:02:16.945999 135477762000704 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:02:28.038166 135477762000704 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 02:02:28.047311 135477762000704 metric_logger.py:289] number parameters: 1.104 billion I0417 02:02:39.781114 135477762000704 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0417 02:02:39.932226 135477762000704 checkpointing.py:798] Waited 0.15110063552856445 seconds for step 10 to finish before starting checkpointing. I0417 02:02:39.934912 135477762000704 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:02:39.936947 135477762000704 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0417 02:02:39.938359 135477762000704 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10. I0417 02:02:40.667928 135477762000704 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:02:40.684277 135477762000704 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:02:40.684398 135477762000704 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 02:02:40.913878 135352790537984 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10 I0417 02:02:40.960876 135477762000704 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278089s I0417 02:02:40.961686 135477762000704 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.384 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2865128517150879 s) (per-host) I0417 02:02:40.961759 135477762000704 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286606s (batch_requests_ready=0.002523s, total_serialization_initiated=0.283355s, others=0.000728s) I0417 02:02:40.961860 135477762000704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290581s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.290564) I0417 02:02:40.962604 135477762000704 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10. I0417 02:02:40.962887 135352798930688 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:22:40.962856 I0417 02:02:40.967466 135477762000704 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:02:40.967757 135352224302848 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:02:40.967897 135477762000704 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391359.9348934, 'wait_for_prev_duration_secs': 5.9604644775390625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776391359.9369848, 'checkpointer_blocking_duration_secs': 1.0260357856750488, 'get_old_steps_start_time': 1776391360.963044, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776391359.9325805, 'checkpoint_manager_blocking_duration_secs': 1.0352811813354492} I0417 02:02:40.968061 135477762000704 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0417 02:02:40.968130 135477762000704 max_utils.py:750] Memstats: After params initialized: I0417 02:02:40.968177 135477762000704 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 02:02:40.968208 135477762000704 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 02:02:40.968234 135477762000704 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 02:02:40.968258 135477762000704 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 02:02:41.328023 135477762000704 metric_logger.py:185] completed step: 10, seconds: 11.416, TFLOP/s/device: 1.190, Tokens/s/device: 179.401, total_weights: 65536, loss: 6.827 I0417 02:02:41.469250 135477762000704 metric_logger.py:185] completed step: 11, seconds: 1.544, TFLOP/s/device: 8.802, Tokens/s/device: 1326.679, total_weights: 65536, loss: 6.522 I0417 02:02:41.617336 135477762000704 metric_logger.py:185] completed step: 12, seconds: 0.014, TFLOP/s/device: 951.346, Tokens/s/device: 143397.283, total_weights: 65536, loss: 6.260 I0417 02:02:41.764820 135477762000704 metric_logger.py:185] completed step: 13, seconds: 0.141, TFLOP/s/device: 96.061, Tokens/s/device: 14479.331, total_weights: 65536, loss: 6.039 I0417 02:02:41.777577 135477762000704 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0417 02:02:42.059961 135477762000704 checkpointing.py:798] Waited 0.282351016998291 seconds for step 15 to finish before starting checkpointing. I0417 02:02:42.062867 135477762000704 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:02:42.661762 135352773752576 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/items I0417 02:02:43.311223 135351692318464 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776391362411905479, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776391362411905479, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 02:02:45.978519 135352252892928 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_0 I0417 02:03:24.582099 135352232695552 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.825735s (commit=39.615342s, array_metadata_write=1.210393s) I0417 02:03:24.583382 135352798930688 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.975 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.90820503234863 s) (per-host) I0417 02:03:24.583480 135352798930688 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.620492s. I0417 02:03:26.521728 135351692318464 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776391362411905479, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/_CHECKPOINT_METADATA I0417 02:03:26.696369 135352798930688 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:03:26.869816 135352798930688 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000753s. I0417 02:03:27.653241 135352798930688 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.214089s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/items I0417 02:03:27.655966 135352798930688 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10/items I0417 02:03:28.221658 135352798930688 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10 I0417 02:03:29.650109 135352798930688 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10`. I0417 02:03:29.651276 135352798930688 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.71 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/10 I0417 02:03:29.652874 135352798930688 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.689882s. I0417 02:03:29.653223 135352224302848 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:03:29.653386 135352224302848 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:03:29.653504 135352224302848 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:03:29.654930 135352224302848 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0417 02:03:29.655258 135477762000704 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 02:03:29.655435 135477762000704 checkpoint_manager.py:1452] Waiting for previous save to complete took 47.592592 seconds. If this number is high, consider checkpointing less frequently. I0417 02:03:29.657149 135477762000704 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 15 I0417 02:03:29.659575 135477762000704 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15. I0417 02:03:30.410960 135477762000704 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:03:30.411102 135477762000704 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 02:03:30.440697 135477762000704 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.032703s I0417 02:03:30.441673 135477762000704 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 29.945 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.051514387130737305 s) (per-host) I0417 02:03:30.441782 135477762000704 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.051647s (batch_requests_ready=0.006147s, total_serialization_initiated=0.044642s, others=0.000857s) I0417 02:03:30.441935 135477762000704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.056464s (all_items=0.000022s, per_item={'items': '0.00002193'}, temp_paths=0.056442) I0417 02:03:30.442958 135477762000704 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.79 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15. I0417 02:03:30.443365 135352224302848 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:23:30.443324 I0417 02:03:30.445852 135477762000704 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:03:30.446183 135351683925760 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:03:30.446386 135477762000704 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391362.0627995, 'wait_for_prev_duration_secs': 47.592591524124146, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776391409.6572201, 'checkpointer_blocking_duration_secs': 0.7863302230834961, 'get_old_steps_start_time': 1776391410.4435792, 'get_old_steps_duration_secs': 4.363059997558594e-05, 'checkpoint_manager_blocking_start_time': 1776391362.0604575, 'checkpoint_manager_blocking_duration_secs': 48.385875940322876} I0417 02:03:30.446539 135477762000704 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0417 02:03:30.626693 135352798930688 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15 I0417 02:03:32.449748 135352773752576 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15/items I0417 02:03:35.824052 135352261285632 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_0 I0417 02:03:52.658765 135477762000704 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 02:03:52.670741 135477762000704 metric_logger.py:185] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.858, Tokens/s/device: 13845.883, total_weights: 65536, loss: 5.713 I0417 02:03:52.813621 135477762000704 metric_logger.py:185] completed step: 16, seconds: 70.893, TFLOP/s/device: 0.192, Tokens/s/device: 28.889, total_weights: 65536, loss: 5.599 I0417 02:03:52.961132 135477762000704 metric_logger.py:185] completed step: 17, seconds: 0.008, TFLOP/s/device: 1599.803, Tokens/s/device: 241139.762, total_weights: 65536, loss: 5.511 I0417 02:03:53.109063 135477762000704 metric_logger.py:185] completed step: 18, seconds: 0.145, TFLOP/s/device: 93.877, Tokens/s/device: 14150.194, total_weights: 65536, loss: 5.442 I0417 02:03:53.257604 135477762000704 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0417 02:03:53.258348 135477762000704 checkpointing.py:798] Waited 0.0007560253143310547 seconds for step 19 to finish before starting checkpointing. I0417 02:03:53.260347 135477762000704 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:04:15.403593 135352232695552 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.762576s (commit=40.061957s, array_metadata_write=1.700619s) I0417 02:04:15.404675 135352224302848 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.091 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.01452445983887 s) (per-host) I0417 02:04:15.404743 135352224302848 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.961234s. I0417 02:04:17.431320 135352224302848 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:04:17.591593 135352224302848 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000402s. I0417 02:04:18.515717 135352224302848 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.933591s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15/items I0417 02:04:18.517386 135352224302848 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15/items I0417 02:04:19.049605 135352224302848 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15 I0417 02:04:21.363685 135352224302848 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15`. I0417 02:04:21.364478 135352224302848 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.71 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/15 I0417 02:04:21.365944 135352224302848 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.922431s. I0417 02:04:21.366067 135351683925760 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:04:21.366153 135351683925760 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:04:21.366239 135351683925760 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:04:21.367969 135351683925760 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0417 02:04:21.368131 135477762000704 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 02:04:21.368258 135477762000704 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.107917 seconds. If this number is high, consider checkpointing less frequently. I0417 02:04:21.369806 135477762000704 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0417 02:04:21.371828 135477762000704 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19. I0417 02:04:22.098888 135477762000704 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:04:22.098987 135477762000704 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 02:04:22.125243 135477762000704 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027856s I0417 02:04:22.126054 135477762000704 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 42.354 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03642082214355469 s) (per-host) I0417 02:04:22.126131 135477762000704 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.036518s (batch_requests_ready=0.002666s, total_serialization_initiated=0.033122s, others=0.000731s) I0417 02:04:22.126223 135477762000704 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041191s (all_items=0.000022s, per_item={'items': '0.00002170'}, temp_paths=0.041169) I0417 02:04:22.126960 135477762000704 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19. I0417 02:04:22.127297 135352782145280 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:24:22.127256 I0417 02:04:22.129258 135477762000704 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:04:22.129502 135351683925760 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:04:22.129634 135477762000704 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391433.2603128, 'wait_for_prev_duration_secs': 28.107916593551636, 'time_between_consecutive_saves_sec': 23.605312824249268, 'checkpointer_blocking_start_time': 1776391461.369865, 'checkpointer_blocking_duration_secs': 0.7575759887695312, 'get_old_steps_start_time': 1776391462.1274598, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776391433.2585797, 'checkpoint_manager_blocking_duration_secs': 28.87102246284485} I0417 02:04:22.129739 135477762000704 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0417 02:04:22.129781 135477762000704 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:04:22.333689 135352224302848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19 I0417 02:04:24.113401 135352773752576 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19/items I0417 02:04:28.753832 135352798930688 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_0 I0417 02:05:07.452591 135352790537984 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.784904s (commit=40.527395s, array_metadata_write=1.257509s) I0417 02:05:07.453787 135352782145280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.820 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.364126682281494 s) (per-host) I0417 02:05:07.453927 135352782145280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.326511s. I0417 02:05:09.413698 135352782145280 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000412s. I0417 02:05:09.636298 135352782145280 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:05:10.581103 135352782145280 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.226145s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19/items I0417 02:05:10.582888 135352782145280 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19/items I0417 02:05:11.132652 135352782145280 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19 I0417 02:05:12.557430 135352782145280 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19`. I0417 02:05:12.559967 135352782145280 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.19 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_true/checkpoints/19 I0417 02:05:12.561632 135352782145280 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.434213s. I0417 02:05:12.561780 135351683925760 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:05:12.561886 135351683925760 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:05:12.561940 135351683925760 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:05:12.563452 135351683925760 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0417 02:05:12.563624 135477762000704 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 02:05:12.563787 135477762000704 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:05:12.564440 135477762000704 metric_logger.py:185] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.314, Tokens/s/device: 13914.556, 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 02:05:23 UTC 2026 EXIT_CODE=0