XPK Start: Fri Apr 17 02:13:55 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 02:14:19.666819: 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 02:14:19.845167 138160378455872 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:14:28.885553 138160378455872 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:14:28.887787 138160378455872 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-faur1-slice-job-0-0.mt-16-zarr3-true-sv-faur1:8482 I0417 02:14:30.162491 138160378455872 max_utils.py:284] Jax distributed system initialized! I0417 02:14:36.053382 138160378455872 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:14:36.053485 138160378455872 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:14:36.053525 138160378455872 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 02:14:36.053559 138160378455872 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:14:36.720144 138160378455872 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:14:36.832952 138160378455872 checkpointing.py:688] Setting up checkpoint logger... I0417 02:14:36.833067 138160378455872 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 02:14:36.833111 138160378455872 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:14:36.833316 138160378455872 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 0x7da775513620>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:14:39.769948 138160378455872 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:14:39.770177 138160378455872 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 0x7d92644bcec0>}, handler_registry=None I0417 02:14:39.770408 138160378455872 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d92644bcec0>` 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 02:14:39.770455 138160378455872 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d9264411be0>` 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 02:14:39.770489 138160378455872 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 0x7d92644bcec0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d92644bcec0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d9264411be0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d9264411be0>}). I0417 02:14:39.770816 138160378455872 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:14:39.770899 138160378455872 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d92386ba020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:14:40.889256 138160378455872 checkpoint_manager.py:1812] Found 0 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_16_zarr3_true/checkpoints I0417 02:14:41.351088 138160378455872 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_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d923864a270> I0417 02:14:41.351255 138160378455872 checkpointing.py:302] Checkpoint manager created! I0417 02:14:42.294695 138160378455872 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:14:42.294818 138160378455872 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 02:14:43.652012 138160378455872 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:14:43.652125 138160378455872 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:14:43.652177 138160378455872 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:14:43.652219 138160378455872 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:14:43.652256 138160378455872 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:14:43.652289 138160378455872 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:14:43.652325 138160378455872 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:14:43.652360 138160378455872 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:14:43.652394 138160378455872 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:14:43.652427 138160378455872 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:14:43.652459 138160378455872 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:14:43.652491 138160378455872 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:14:44.063396 138160378455872 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:14:44.063485 138160378455872 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:14:44.068728 138160378455872 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:14:44.068784 138160378455872 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:14:44.084856 138160378455872 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:14:44.084914 138160378455872 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:14:44.100308 138160378455872 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:14:44.100367 138160378455872 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:14:44.123983 138160378455872 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 02:14:44.124050 138160378455872 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:14:44.139612 138160378455872 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 02:14:44.139669 138160378455872 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:14:44.155059 138160378455872 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 02:14:44.155116 138160378455872 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:14:44.189699 138160378455872 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:14:44.189805 138160378455872 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:14:44.210026 138160378455872 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:14:44.210089 138160378455872 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:14:55.094324 138160378455872 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 02:14:55.103316 138160378455872 metric_logger.py:289] number parameters: 1.104 billion I0417 02:15:06.689239 138160378455872 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 02:15:06.841981 138160378455872 checkpointing.py:798] Waited 0.1527252197265625 seconds for step 0 to finish before starting checkpointing. I0417 02:15:06.844627 138160378455872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:15:06.846366 138160378455872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 02:15:06.847693 138160378455872 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_16_zarr3_true/checkpoints/0. I0417 02:15:07.541987 138160378455872 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:15:07.561272 138160378455872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:15:07.561342 138160378455872 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 02:15:07.837502 138160378455872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277608s I0417 02:15:07.838270 138160378455872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.346 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2885410785675049 s) (per-host) I0417 02:15:07.838331 138160378455872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288620s (batch_requests_ready=0.003285s, total_serialization_initiated=0.284661s, others=0.000674s) I0417 02:15:07.838407 138160378455872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292446s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.292429) I0417 02:15:07.839210 138160378455872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.99 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_16_zarr3_true/checkpoints/0. I0417 02:15:07.839591 138032879064832 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:35:07.839557 I0417 02:15:07.844805 138160378455872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:15:07.845127 138030746818304 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:15:07.845282 138160378455872 standard_logger.py:34] {'step': 0, '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_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392106.844607, 'wait_for_prev_duration_secs': 6.175041198730469e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392106.8464046, 'checkpointer_blocking_duration_secs': 0.9933328628540039, 'get_old_steps_start_time': 1776392107.8397622, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776392106.8424065, 'checkpoint_manager_blocking_duration_secs': 1.0028388500213623} I0417 02:15:07.845389 138160378455872 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 02:15:07.845455 138160378455872 max_utils.py:750] Memstats: After params initialized: I0417 02:15:07.845503 138160378455872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 02:15:07.845535 138160378455872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 02:15:07.845561 138160378455872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 02:15:07.845587 138160378455872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 02:15:07.862906 138032355890944 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_16_zarr3_true/checkpoints/0 I0417 02:15:08.275407 138160378455872 metric_logger.py:185] completed step: 0, seconds: 11.283, TFLOP/s/device: 1.204, Tokens/s/device: 181.516, total_weights: 65536, loss: 10.874 I0417 02:15:08.278548 138160378455872 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/tensorboard/' I0417 02:15:08.736824 138160378455872 metric_logger.py:185] completed step: 1, seconds: 1.583, TFLOP/s/device: 8.585, Tokens/s/device: 1293.972, total_weights: 65536, loss: 10.874 I0417 02:15:08.878548 138160378455872 metric_logger.py:185] completed step: 2, seconds: 0.462, TFLOP/s/device: 29.380, Tokens/s/device: 4428.491, total_weights: 65536, loss: 10.560 I0417 02:15:09.026181 138160378455872 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1122.996, Tokens/s/device: 169270.188, total_weights: 65536, loss: 9.979 I0417 02:15:09.038174 138160378455872 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0417 02:15:09.321739 138160378455872 checkpointing.py:798] Waited 0.28353214263916016 seconds for step 5 to finish before starting checkpointing. I0417 02:15:09.324560 138160378455872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:15:10.038207 138032339105536 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_16_zarr3_true/checkpoints/0/items I0417 02:15:10.054954 138029683435264 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392109615182700, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776392109615182700, "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_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0417 02:15:11.167496 2769 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:15:13.387296 138031804315392 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_16_zarr3_true/checkpoints/0/items/array_metadatas/process_0 I0417 02:15:46.543397 138030786541312 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.395469s (commit=34.094959s, array_metadata_write=1.300510s) I0417 02:15:46.544913 138032879064832 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 40.508 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.99512076377869 s) (per-host) I0417 02:15:46.545078 138032879064832 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.705375s. I0417 02:15:47.247549 138032879064832 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392109615182700, 'commit_timestamp_nsecs': None, '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_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0417 02:15:48.202608 138029683435264 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776392109615182700, '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_16_zarr3_true/checkpoints/0/_CHECKPOINT_METADATA I0417 02:15:48.427416 138032879064832 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:15:48.559692 138032879064832 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000733s. I0417 02:15:49.079973 138032879064832 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.513632s. 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_16_zarr3_true/checkpoints/0/items I0417 02:15:49.082151 138032879064832 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_16_zarr3_true/checkpoints/0/items I0417 02:15:49.638550 138032879064832 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_16_zarr3_true/checkpoints/0 I0417 02:15:51.041001 138032879064832 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_16_zarr3_true/checkpoints/0`. I0417 02:15:51.042114 138032879064832 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 44.20 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/0 I0417 02:15:51.043514 138032879064832 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 43.203808s. I0417 02:15:51.043815 138030746818304 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:15:51.043998 138030746818304 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:15:51.044108 138030746818304 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:15:51.046672 138030746818304 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 02:15:51.046960 138160378455872 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0417 02:15:51.047084 138160378455872 checkpoint_manager.py:1452] Waiting for previous save to complete took 41.722533 seconds. If this number is high, consider checkpointing less frequently. I0417 02:15:51.049015 138160378455872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0417 02:15:51.051308 138160378455872 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_16_zarr3_true/checkpoints/5. I0417 02:15:51.358408 138160378455872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:15:51.358544 138160378455872 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 02:15:51.400758 138160378455872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045331s I0417 02:15:51.401745 138160378455872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 23.725 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0650186538696289 s) (per-host) I0417 02:15:51.401863 138160378455872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.065149s (batch_requests_ready=0.006799s, total_serialization_initiated=0.057469s, others=0.000880s) I0417 02:15:51.402000 138160378455872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.069891s (all_items=0.000022s, per_item={'items': '0.00002241'}, temp_paths=0.069869) I0417 02:15:51.402968 138160378455872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/5. I0417 02:15:51.403407 138030746818304 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:35:51.403364 I0417 02:15:51.405756 138160378455872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:15:51.406148 138031306626816 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:15:51.406396 138160378455872 standard_logger.py:34] {'step': 5, '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_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392109.3245134, 'wait_for_prev_duration_secs': 41.722532749176025, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776392151.049087, 'checkpointer_blocking_duration_secs': 0.3545036315917969, 'get_old_steps_start_time': 1776392151.4036186, 'get_old_steps_duration_secs': 4.291534423828125e-05, 'checkpoint_manager_blocking_start_time': 1776392109.322254, 'checkpoint_manager_blocking_duration_secs': 42.084089517593384} I0417 02:15:51.406543 138160378455872 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0417 02:15:52.032538 138032879064832 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_16_zarr3_true/checkpoints/5 I0417 02:15:53.767545 138031835105024 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_16_zarr3_true/checkpoints/5/items I0417 02:15:57.947758 138031804315392 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_16_zarr3_true/checkpoints/5/items/array_metadatas/process_0 I0417 02:16:15.278959 138160378455872 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 95.051, Tokens/s/device: 14327.189, total_weights: 65536, loss: 9.457 I0417 02:16:15.290040 138160378455872 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.379, Tokens/s/device: 13924.395, total_weights: 65536, loss: 8.955 I0417 02:16:15.434049 138160378455872 metric_logger.py:185] completed step: 6, seconds: 66.252, TFLOP/s/device: 0.205, Tokens/s/device: 30.912, total_weights: 65536, loss: 8.465 I0417 02:16:15.581515 138160378455872 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1599.238, Tokens/s/device: 241054.614, total_weights: 65536, loss: 7.998 I0417 02:16:15.729052 138160378455872 metric_logger.py:185] completed step: 8, seconds: 0.147, TFLOP/s/device: 92.576, Tokens/s/device: 13954.090, total_weights: 65536, loss: 7.567 I0417 02:16:15.877601 138160378455872 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 02:16:15.878296 138160378455872 checkpointing.py:798] Waited 0.0007066726684570312 seconds for step 9 to finish before starting checkpointing. I0417 02:16:15.880181 138160378455872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:16:36.305655 138031315019520 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.558916s (commit=39.357887s, array_metadata_write=1.201030s) I0417 02:16:36.307735 138030746818304 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.125 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.970993757247925 s) (per-host) I0417 02:16:36.307880 138030746818304 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.904328s. I0417 02:16:37.741063 138030746818304 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000385s. I0417 02:16:37.938299 138030746818304 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:16:38.583462 138030746818304 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.485177s. 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_16_zarr3_true/checkpoints/5/items I0417 02:16:38.585938 138030746818304 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_16_zarr3_true/checkpoints/5/items I0417 02:16:39.197428 138030746818304 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_16_zarr3_true/checkpoints/5 I0417 02:16:40.635361 138030746818304 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_16_zarr3_true/checkpoints/5`. I0417 02:16:40.636229 138030746818304 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.59 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/5 I0417 02:16:40.637692 138030746818304 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.234138s. I0417 02:16:40.637907 138031306626816 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:16:40.638032 138031306626816 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:16:40.638104 138031306626816 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:16:40.640314 138031306626816 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 02:16:40.640488 138160378455872 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0417 02:16:40.640608 138160378455872 checkpoint_manager.py:1452] Waiting for previous save to complete took 24.760435 seconds. If this number is high, consider checkpointing less frequently. I0417 02:16:40.642084 138160378455872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 02:16:40.644063 138160378455872 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_16_zarr3_true/checkpoints/9. I0417 02:16:40.935084 138160378455872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:16:40.935177 138160378455872 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 02:16:40.961532 138160378455872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027921s I0417 02:16:40.962342 138160378455872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 42.253 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03650784492492676 s) (per-host) I0417 02:16:40.962422 138160378455872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.036607s (batch_requests_ready=0.002760s, total_serialization_initiated=0.033124s, others=0.000723s) I0417 02:16:40.962514 138160378455872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040520s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.040505) I0417 02:16:40.963155 138160378455872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.32 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_16_zarr3_true/checkpoints/9. I0417 02:16:40.963437 138031804315392 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:36:40.963398 I0417 02:16:40.970885 138160378455872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:16:40.971113 138031306626816 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:16:40.971250 138160378455872 standard_logger.py:34] {'step': 9, '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_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776392175.880147, 'wait_for_prev_duration_secs': 24.760434865951538, 'time_between_consecutive_saves_sec': 24.83340287208557, 'checkpointer_blocking_start_time': 1776392200.642125, 'checkpointer_blocking_duration_secs': 0.3214454650878906, 'get_old_steps_start_time': 1776392200.963587, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776392175.8785193, 'checkpoint_manager_blocking_duration_secs': 25.092700719833374} I0417 02:16:40.971359 138160378455872 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 02:16:40.971400 138160378455872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:16:41.169755 138030746818304 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_16_zarr3_true/checkpoints/9 I0417 02:16:43.786869 138032355890944 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_16_zarr3_true/checkpoints/9/items I0417 02:16:46.692998 138031835592448 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_16_zarr3_true/checkpoints/9/items/array_metadatas/process_0 I0417 02:17:25.395675 138031819159296 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.486737s (commit=39.282631s, array_metadata_write=1.204106s) I0417 02:17:25.396890 138031804315392 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.520 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.47101545333862 s) (per-host) I0417 02:17:25.397022 138031804315392 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.433475s. I0417 02:17:27.188172 138031804315392 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 02:17:27.210081 138031804315392 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000431s. I0417 02:17:27.818586 138031804315392 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.587068s. 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_16_zarr3_true/checkpoints/9/items I0417 02:17:27.820222 138031804315392 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_16_zarr3_true/checkpoints/9/items I0417 02:17:28.373537 138031804315392 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_16_zarr3_true/checkpoints/9 I0417 02:17:29.807339 138031804315392 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_16_zarr3_true/checkpoints/9`. I0417 02:17:29.808119 138031804315392 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.17 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_16_zarr3_true/checkpoints/9 I0417 02:17:29.809598 138031804315392 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.846048s. I0417 02:17:29.809805 138031306626816 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:17:29.809931 138031306626816 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:17:29.809981 138031306626816 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:17:29.812319 138031306626816 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 02:17:29.812458 138160378455872 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 02:17:29.812566 138160378455872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:17:29.813270 138160378455872 metric_logger.py:185] completed step: 9, seconds: 0.146, TFLOP/s/device: 92.938, Tokens/s/device: 14008.591, total_weights: 65536, loss: 7.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 02:17:41 UTC 2026 EXIT_CODE=0