XPK Start: Wed Apr 22 15:27:00 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-22 15:27:24.561587: 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) I0422 15:27:24.741492 134339425863488 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 15:27:33.782659 134339425863488 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 15:27:33.784997 134339425863488 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-hnbjc-slice-job-0-0.mt-16-zarr3-true-sv-hnbjc:8482 I0422 15:27:35.336152 134339425863488 max_utils.py:284] Jax distributed system initialized! I0422 15:27:41.454684 134339425863488 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 15:27:41.454785 134339425863488 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 15:27:41.454823 134339425863488 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 I0422 15:27:41.454857 134339425863488 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 15:27:42.178369 134339425863488 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 15:27:42.291892 134339425863488 checkpointing.py:688] Setting up checkpoint logger... I0422 15:27:42.292008 134339425863488 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 15:27:42.292052 134339425863488 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 15:27:42.292261 134339425863488 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 0x7a2dba518f50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 15:27:45.220860 134339425863488 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 15:27:45.221105 134339425863488 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a183c539850>}, handler_registry=None I0422 15:27:45.221351 134339425863488 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a183c539850>` 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`. I0422 15:27:45.221397 134339425863488 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a183c550bf0>` 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`. I0422 15:27:45.221433 134339425863488 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 0x7a183c539850>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a183c539850>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a183c550bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a183c550bf0>}). I0422 15:27:45.221746 134339425863488 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 15:27:45.221817 134339425863488 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a183c51dda0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 15:27:45.973939 134339425863488 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints I0422 15:27:46.084949 134339425863488 checkpoint_manager.py:929] [process=6][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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a183c3ca840> I0422 15:27:46.085162 134339425863488 checkpointing.py:302] Checkpoint manager created! I0422 15:27:47.052743 134339425863488 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 15:27:47.052855 134339425863488 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0422 15:27:48.416146 134339425863488 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0422 15:27:48.416260 134339425863488 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:27:48.416311 134339425863488 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:27:48.416352 134339425863488 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0422 15:27:48.416389 134339425863488 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:27:48.416422 134339425863488 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:27:48.416456 134339425863488 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:27:48.416496 134339425863488 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0422 15:27:48.416530 134339425863488 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:27:48.416563 134339425863488 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:27:48.416595 134339425863488 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0422 15:27:48.416625 134339425863488 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0422 15:27:48.813177 134339425863488 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:27:48.813266 134339425863488 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:27:48.818668 134339425863488 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:27:48.818728 134339425863488 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:27:48.834998 134339425863488 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:27:48.835056 134339425863488 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:27:48.850689 134339425863488 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:27:48.850746 134339425863488 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:27:48.875761 134339425863488 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:27:48.875828 134339425863488 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:27:48.891678 134339425863488 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:27:48.891740 134339425863488 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:27:48.907489 134339425863488 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:27:48.907548 134339425863488 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:27:48.943731 134339425863488 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 15:27:48.943836 134339425863488 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:27:48.964299 134339425863488 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 15:27:48.964361 134339425863488 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 15:28:00.007769 134339425863488 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. I0422 15:28:00.010072 134339425863488 metric_logger.py:301] number parameters: 1.104 billion I0422 15:28:11.426648 134339425863488 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0422 15:28:11.579632 134339425863488 checkpointing.py:798] Waited 0.152968168258667 seconds for step 0 to finish before starting checkpointing. I0422 15:28:11.581994 134339425863488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:28:11.583837 134339425863488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0422 15:28:11.585345 134339425863488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/0. I0422 15:28:11.885842 134339425863488 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 15:28:11.886778 134339425863488 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:28:11.886834 134339425863488 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 I0422 15:28:12.163491 134339425863488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278410s I0422 15:28:12.163671 134339425863488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.365 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2875032424926758 s) (per-host) I0422 15:28:12.163727 134339425863488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287570s (batch_requests_ready=0.003155s, total_serialization_initiated=0.284340s, others=0.000075s) I0422 15:28:12.163830 134339425863488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291642s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.291623) I0422 15:28:12.164633 134339425863488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.58 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/0. I0422 15:28:12.164969 134210315466496 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:48:12.164937 I0422 15:28:12.172953 134339425863488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:28:12.173261 134208735409920 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:28:12.173420 134339425863488 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776871691.5819747, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776871691.5838766, 'checkpointer_blocking_duration_secs': 0.5812509059906006, 'get_old_steps_start_time': 1776871692.1651552, 'get_old_steps_duration_secs': 3.24249267578125e-05, 'checkpoint_manager_blocking_start_time': 1776871691.5800302, 'checkpoint_manager_blocking_duration_secs': 0.593350887298584} I0422 15:28:12.173534 134339425863488 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0422 15:28:12.173601 134339425863488 max_utils.py:750] Memstats: After params initialized: I0422 15:28:12.173648 134339425863488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0422 15:28:12.173680 134339425863488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0422 15:28:12.173706 134339425863488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0422 15:28:12.173730 134339425863488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0422 15:28:12.493598 134339425863488 metric_logger.py:196] completed step: 0, seconds: 11.415, TFLOP/s/device: 1.190, Tokens/s/device: 179.406, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0422 15:28:12.669061 134339425863488 metric_logger.py:196] completed step: 1, seconds: 1.065, TFLOP/s/device: 12.757, Tokens/s/device: 1922.920, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0422 15:28:13.096501 134339425863488 metric_logger.py:196] completed step: 2, seconds: 0.030, TFLOP/s/device: 460.549, Tokens/s/device: 69419.022, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0422 15:28:13.243825 134339425863488 metric_logger.py:196] completed step: 3, seconds: 0.417, TFLOP/s/device: 32.552, Tokens/s/device: 4906.647, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0422 15:28:13.249637 134339425863488 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0422 15:28:13.538776 134339425863488 checkpointing.py:798] Waited 0.2891097068786621 seconds for step 5 to finish before starting checkpointing. I0422 15:28:13.541525 134339425863488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:28:15.855669 2835 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 15:28:17.897222 134209750587136 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/0/items/array_metadatas/process_6 I0422 15:28:49.591167 134210315466496 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 41.883 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.71494460105896 s) (per-host) I0422 15:28:49.591300 134210315466496 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.426215s. I0422 15:28:57.982071 134210315466496 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 45.816978s. I0422 15:28:57.982372 134208735409920 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:28:57.982495 134208735409920 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:28:57.982557 134208735409920 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:28:57.984668 134208735409920 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 15:28:57.984867 134339425863488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0422 15:28:57.985015 134339425863488 checkpoint_manager.py:1452] Waiting for previous save to complete took 44.443488 seconds. If this number is high, consider checkpointing less frequently. I0422 15:28:57.986896 134339425863488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0422 15:28:57.989337 134339425863488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/5. I0422 15:28:58.714486 134339425863488 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:28:58.714583 134339425863488 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 I0422 15:28:58.748640 134339425863488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035715s I0422 15:28:58.748812 134339425863488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 35.812 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04307413101196289 s) (per-host) I0422 15:28:58.748863 134339425863488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.043136s (batch_requests_ready=0.002755s, total_serialization_initiated=0.040311s, others=0.000070s) I0422 15:28:58.748957 134339425863488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.048127s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.048111) I0422 15:28:58.749743 134339425863488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/5. I0422 15:28:58.750042 134208735409920 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:48:58.750010 I0422 15:28:58.771209 134339425863488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:28:58.771480 134209256322816 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:28:58.771626 134339425863488 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776871693.5414958, 'wait_for_prev_duration_secs': 44.44348764419556, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776871737.9869344, 'checkpointer_blocking_duration_secs': 0.7632176876068115, 'get_old_steps_start_time': 1776871738.7501767, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776871693.539137, 'checkpoint_manager_blocking_duration_secs': 45.23245191574097} I0422 15:28:58.771736 134339425863488 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0422 15:28:58.772844 134339425863488 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.846, Tokens/s/device: 12487.424, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0422 15:28:58.780689 134339425863488 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.418, Tokens/s/device: 13930.172, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0422 15:29:04.079180 134209750587136 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/5/items/array_metadatas/process_6 I0422 15:29:22.509766 134339425863488 metric_logger.py:196] completed step: 6, seconds: 45.529, TFLOP/s/device: 0.298, Tokens/s/device: 44.982, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0422 15:29:22.657783 134339425863488 metric_logger.py:196] completed step: 7, seconds: 23.584, TFLOP/s/device: 0.576, Tokens/s/device: 86.839, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0422 15:29:22.805105 134339425863488 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.777, Tokens/s/device: 13381.423, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0422 15:29:22.953372 134339425863488 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0422 15:29:22.954045 134339425863488 checkpointing.py:798] Waited 0.0006854534149169922 seconds for step 9 to finish before starting checkpointing. I0422 15:29:22.956182 134339425863488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:29:40.527454 134208735409920 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.770 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.82169461250305 s) (per-host) I0422 15:29:40.527560 134208735409920 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.776886s. I0422 15:29:48.907710 134208735409920 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.157020s. I0422 15:29:48.907926 134209256322816 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:29:48.907985 134209256322816 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:29:48.908038 134209256322816 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:29:48.910299 134209256322816 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0422 15:29:48.910467 134339425863488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0422 15:29:48.910594 134339425863488 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.954417 seconds. If this number is high, consider checkpointing less frequently. I0422 15:29:48.912090 134339425863488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0422 15:29:48.914181 134339425863488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/9. I0422 15:29:49.634906 134339425863488 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:29:49.634998 134339425863488 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 I0422 15:29:49.670449 134339425863488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037160s I0422 15:29:49.670619 134339425863488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.925 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04416799545288086 s) (per-host) I0422 15:29:49.670668 134339425863488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044228s (batch_requests_ready=0.002879s, total_serialization_initiated=0.041284s, others=0.000065s) I0422 15:29:49.670763 134339425863488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.048349s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.048337) I0422 15:29:49.671482 134339425863488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/9. I0422 15:29:49.671823 134209256322816 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:49:49.671786 I0422 15:29:49.673798 134339425863488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:29:49.674018 134206034269952 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:29:49.674156 134339425863488 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776871762.956153, 'wait_for_prev_duration_secs': 25.95441722869873, 'time_between_consecutive_saves_sec': 24.97144651412964, 'checkpointer_blocking_start_time': 1776871788.9121401, 'checkpointer_blocking_duration_secs': 0.7598273754119873, 'get_old_steps_start_time': 1776871789.6719856, 'get_old_steps_duration_secs': 2.4080276489257812e-05, 'checkpoint_manager_blocking_start_time': 1776871762.9542933, 'checkpoint_manager_blocking_duration_secs': 26.71983242034912} I0422 15:29:49.674330 134339425863488 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0422 15:29:49.674375 134339425863488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:29:54.858190 134209750587136 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_16_zarr3_true/checkpoints/9/items/array_metadatas/process_6 I0422 15:30:31.737137 134209256322816 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.511 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.11062932014465 s) (per-host) I0422 15:30:31.737268 134209256322816 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.065336s. I0422 15:30:38.873451 134209256322816 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.201502s. I0422 15:30:38.873737 134206034269952 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:30:38.873859 134206034269952 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:30:38.873904 134206034269952 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:30:38.875509 134206034269952 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 15:30:38.875677 134339425863488 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0422 15:30:38.875836 134339425863488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:30:38.876935 134339425863488 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.568, Tokens/s/device: 13802.120, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Wed Apr 22 15:30:50 UTC 2026 EXIT_CODE=0