XPK Start: Fri Apr 17 04:38:52 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 04:39:16.430121: 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 04:39:16.608587 132571305051968 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 04:39:25.649945 132571305051968 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 04:39:25.652255 132571305051968 distributed.py:172] Connecting to JAX distributed service on mt-01-base-bq0kj-slice-job-0-0.mt-01-base-bq0kj:8482 I0417 04:39:26.842730 132571305051968 max_utils.py:284] Jax distributed system initialized! I0417 04:39:32.876860 132571305051968 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 04:39:32.876963 132571305051968 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 04:39:32.877003 132571305051968 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 04:39:32.877037 132571305051968 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 04:39:33.548088 132571305051968 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 04:39:33.661496 132571305051968 checkpointing.py:688] Setting up checkpoint logger... I0417 04:39:33.661612 132571305051968 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 04:39:33.661658 132571305051968 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 04:39:33.661878 132571305051968 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 0x7891f1c4d040>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 04:39:37.343488 132571305051968 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 04:39:37.343717 132571305051968 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 0x787db04d4a10>}, handler_registry=None I0417 04:39:37.343970 132571305051968 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x787db04d4a10>` 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 04:39:37.344019 132571305051968 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787db82e26c0>` 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 04:39:37.344056 132571305051968 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 0x787db04d4a10>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x787db04d4a10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787db82e26c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787db82e26c0>}). I0417 04:39:37.344368 132571305051968 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 04:39:37.344439 132571305051968 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x787db815e020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 04:39:38.795549 132571305051968 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints I0417 04:39:39.654078 132571305051968 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=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7890e61782c0> I0417 04:39:39.654251 132571305051968 checkpointing.py:302] Checkpoint manager created! I0417 04:39:40.599969 132571305051968 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 04:39:40.600090 132571305051968 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 04:39:41.958715 132571305051968 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 04:39:41.958832 132571305051968 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 04:39:41.958904 132571305051968 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 04:39:41.958976 132571305051968 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 04:39:41.959017 132571305051968 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 04:39:41.959052 132571305051968 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 04:39:41.959087 132571305051968 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:39:41.959121 132571305051968 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 04:39:41.959154 132571305051968 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:39:41.959185 132571305051968 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 04:39:41.959215 132571305051968 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 04:39:41.959245 132571305051968 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 04:39:42.362776 132571305051968 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:39:42.362890 132571305051968 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:39:42.368255 132571305051968 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 04:39:42.368312 132571305051968 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:39:42.384509 132571305051968 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:39:42.384572 132571305051968 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:39:42.400127 132571305051968 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 04:39:42.400184 132571305051968 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 04:39:42.424140 132571305051968 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 04:39:42.424209 132571305051968 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:39:42.439912 132571305051968 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 04:39:42.439973 132571305051968 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:39:42.455612 132571305051968 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 04:39:42.455682 132571305051968 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:39:42.485555 132571305051968 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 04:39:42.485625 132571305051968 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 04:39:42.505305 132571305051968 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 04:39:42.505369 132571305051968 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 04:39:53.559054 132571305051968 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 04:39:53.568092 132571305051968 metric_logger.py:289] number parameters: 1.104 billion I0417 04:40:04.938578 132571305051968 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 04:40:05.090934 132571305051968 checkpointing.py:798] Waited 0.15233922004699707 seconds for step 0 to finish before starting checkpointing. I0417 04:40:05.093114 132571305051968 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:40:05.094784 132571305051968 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 04:40:05.096176 132571305051968 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0. I0417 04:40:06.249880 132571305051968 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 04:40:06.269479 132571305051968 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 04:40:06.269549 132571305051968 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 04:40:06.540766 132571305051968 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272865s I0417 04:40:06.541714 132571305051968 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.436 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2837808132171631 s) (per-host) I0417 04:40:06.541795 132571305051968 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283883s (batch_requests_ready=0.003068s, total_serialization_initiated=0.279966s, others=0.000849s) I0417 04:40:06.541903 132571305051968 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287776s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.287760) I0417 04:40:06.542742 132571305051968 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.45 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0. I0417 04:40:06.543083 132442654230272 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:00:06.543046 I0417 04:40:06.553026 132571305051968 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:40:06.553301 132440505227008 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:40:06.553450 132571305051968 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776400805.0930948, 'wait_for_prev_duration_secs': 5.9604644775390625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776400805.0948303, 'checkpointer_blocking_duration_secs': 1.4484126567840576, 'get_old_steps_start_time': 1776400806.5432673, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776400805.0913274, 'checkpoint_manager_blocking_duration_secs': 1.4620866775512695} I0417 04:40:06.553560 132571305051968 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 04:40:06.553628 132571305051968 max_utils.py:750] Memstats: After params initialized: I0417 04:40:06.553679 132571305051968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 04:40:06.553707 132571305051968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 04:40:06.553733 132571305051968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 04:40:06.553756 132571305051968 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 04:40:06.596617 132442645837568 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0 I0417 04:40:06.892815 132571305051968 metric_logger.py:185] completed step: 0, seconds: 11.042, TFLOP/s/device: 1.230, Tokens/s/device: 185.470, total_weights: 65536, loss: 10.874 I0417 04:40:06.895443 132571305051968 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/tensorboard/' I0417 04:40:07.318004 132571305051968 metric_logger.py:185] completed step: 1, seconds: 1.951, TFLOP/s/device: 6.965, Tokens/s/device: 1049.849, total_weights: 65536, loss: 10.874 I0417 04:40:07.459645 132571305051968 metric_logger.py:185] completed step: 2, seconds: 0.426, TFLOP/s/device: 31.883, Tokens/s/device: 4805.797, total_weights: 65536, loss: 10.262 I0417 04:40:07.607099 132571305051968 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1129.625, Tokens/s/device: 170269.371, total_weights: 65536, loss: 9.732 I0417 04:40:08.424475 132441616410368 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/items I0417 04:40:08.576266 132439435732736 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776400808159051173, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776400808159051173, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 04:40:09.514981 2832 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 04:40:11.622913 132440549676800 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/items/array_metadatas/process_0 I0417 04:40:31.522292 132571305051968 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.917, Tokens/s/device: 14306.971, total_weights: 65536, loss: 9.274 I0417 04:40:31.532402 132571305051968 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.439, Tokens/s/device: 13933.395, total_weights: 65536, loss: 8.888 I0417 04:40:31.676740 132571305051968 metric_logger.py:185] completed step: 6, seconds: 23.914, TFLOP/s/device: 0.568, Tokens/s/device: 85.642, total_weights: 65536, loss: 8.589 I0417 04:40:31.824210 132571305051968 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1713.383, Tokens/s/device: 258259.773, total_weights: 65536, loss: 8.380 I0417 04:40:31.971661 132571305051968 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 92.777, Tokens/s/device: 13984.390, total_weights: 65536, loss: 8.251 I0417 04:40:32.120193 132571305051968 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 04:40:32.120832 132571305051968 checkpointing.py:798] Waited 0.0006580352783203125 seconds for step 9 to finish before starting checkpointing. I0417 04:40:32.122862 132571305051968 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:40:46.319099 132440541284096 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.814137s (commit=35.501832s, array_metadata_write=1.312304s) I0417 04:40:46.320224 132442654230272 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.429 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.062294483184814 s) (per-host) I0417 04:40:46.320308 132442654230272 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.777112s. I0417 04:40:47.006933 132442654230272 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776400808159051173, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 04:40:47.784791 132439435732736 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776400808159051173, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 04:40:47.795506 132442654230272 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000359s. I0417 04:40:48.385064 132442654230272 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:40:49.313506 132442654230272 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.183470s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/items I0417 04:40:49.315147 132442654230272 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0/items I0417 04:40:49.900302 132442654230272 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0 I0417 04:40:51.757566 132442654230272 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0`. I0417 04:40:51.758406 132442654230272 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.66 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/0 I0417 04:40:51.759877 132442654230272 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.216681s. I0417 04:40:51.760094 132440505227008 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:40:51.760223 132440505227008 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:40:51.760287 132440505227008 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:40:51.763451 132440505227008 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 04:40:51.763572 132571305051968 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 04:40:51.763649 132571305051968 checkpoint_manager.py:1452] Waiting for previous save to complete took 19.640808 seconds. If this number is high, consider checkpointing less frequently. I0417 04:40:51.765590 132571305051968 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 04:40:51.767653 132571305051968 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9. I0417 04:40:52.486138 132571305051968 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 04:40:52.486235 132571305051968 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 04:40:52.513479 132571305051968 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028824s I0417 04:40:52.514223 132571305051968 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.922 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03769564628601074 s) (per-host) I0417 04:40:52.514291 132571305051968 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037778s (batch_requests_ready=0.002636s, total_serialization_initiated=0.034449s, others=0.000694s) I0417 04:40:52.514365 132571305051968 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041754s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.041737) I0417 04:40:52.515038 132571305051968 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9. I0417 04:40:52.515295 132440505227008 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 05:00:52.515273 I0417 04:40:52.521439 132571305051968 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 04:40:52.521681 132441582839552 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 04:40:52.521811 132571305051968 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776400832.1228142, 'wait_for_prev_duration_secs': 19.64080834388733, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776400851.7656293, 'checkpointer_blocking_duration_secs': 0.7497615814208984, 'get_old_steps_start_time': 1776400852.5154145, 'get_old_steps_duration_secs': 3.1948089599609375e-05, 'checkpoint_manager_blocking_start_time': 1776400832.1210773, 'checkpoint_manager_blocking_duration_secs': 20.400697231292725} I0417 04:40:52.521945 132571305051968 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 04:40:52.521991 132571305051968 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 04:40:52.721216 132442654230272 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9 I0417 04:40:55.311573 132441079539456 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9/items I0417 04:40:58.159036 132441599624960 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9/items/array_metadatas/process_0 I0417 04:41:39.311759 132441591232256 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.896114s (commit=41.271635s, array_metadata_write=1.624479s) I0417 04:41:39.313014 132440505227008 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.726 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.83643960952759 s) (per-host) I0417 04:41:39.313146 132440505227008 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.797774s. I0417 04:41:41.025606 132440505227008 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 04:41:41.596723 132440505227008 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000397s. I0417 04:41:42.076571 132440505227008 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.889557s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9/items I0417 04:41:42.078211 132440505227008 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9/items I0417 04:41:42.625271 132440505227008 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9 I0417 04:41:44.436245 132440505227008 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9`. I0417 04:41:44.437110 132440505227008 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.67 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_01_base/checkpoints/9 I0417 04:41:44.438654 132440505227008 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.923281s. I0417 04:41:44.438831 132441582839552 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 04:41:44.438953 132441582839552 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 04:41:44.439014 132441582839552 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 04:41:44.440961 132441582839552 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 04:41:44.442737 132571305051968 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 04:41:44.442866 132571305051968 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 04:41:44.443546 132571305051968 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.696, Tokens/s/device: 13972.178, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 04:41:53 UTC 2026 EXIT_CODE=0