XPK Start: Fri Apr 17 06:10:47 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 06:11:11.239766: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0417 06:11:11.419579 138534581335872 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 06:11:20.459824 138534581335872 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 06:11:20.462203 138534581335872 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-863zo-slice-job-0-0.mt-15-ocdbt-false-sv-863zo:8482 I0417 06:11:21.387990 138534581335872 max_utils.py:284] Jax distributed system initialized! I0417 06:11:27.384933 138534581335872 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 06:11:27.385036 138534581335872 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 06:11:27.385082 138534581335872 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0417 06:11:27.385118 138534581335872 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 06:11:28.053464 138534581335872 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 06:11:28.168240 138534581335872 checkpointing.py:688] Setting up checkpoint logger... I0417 06:11:28.168353 138534581335872 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0417 06:11:28.168399 138534581335872 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 06:11:28.168596 138534581335872 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7dfe60b4ce30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 06:11:31.159810 138534581335872 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 06:11:31.160064 138534581335872 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 0x7dea1c6a9a60>}, handler_registry=None I0417 06:11:31.160303 138534581335872 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dea1c6a9a60>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0417 06:11:31.160351 138534581335872 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dea1c5b3f80>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0417 06:11:31.160387 138534581335872 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 0x7dea1c6a9a60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dea1c6a9a60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dea1c5b3f80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dea1c5b3f80>}). I0417 06:11:31.160705 138534581335872 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 06:11:31.160773 138534581335872 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7dea1c586020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 06:11:32.752111 138534581335872 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_15_ocdbt_false/checkpoints I0417 06:11:33.207932 138534581335872 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dea1c137650> I0417 06:11:33.208103 138534581335872 checkpointing.py:302] Checkpoint manager created! I0417 06:11:34.163698 138534581335872 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 06:11:34.163815 138534581335872 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 06:11:35.554159 138534581335872 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 06:11:35.554274 138534581335872 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:11:35.554326 138534581335872 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 06:11:35.554370 138534581335872 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 06:11:35.554406 138534581335872 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:11:35.554440 138534581335872 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 06:11:35.554476 138534581335872 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:11:35.554511 138534581335872 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 06:11:35.554544 138534581335872 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:11:35.554576 138534581335872 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 06:11:35.554607 138534581335872 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 06:11:35.554639 138534581335872 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 06:11:35.959227 138534581335872 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:11:35.959329 138534581335872 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:11:35.964788 138534581335872 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 06:11:35.964861 138534581335872 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:11:35.981479 138534581335872 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:11:35.981540 138534581335872 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:11:35.997358 138534581335872 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 06:11:35.997415 138534581335872 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 06:11:36.021642 138534581335872 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:11:36.021711 138534581335872 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:11:36.037634 138534581335872 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:11:36.037697 138534581335872 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:11:36.053617 138534581335872 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 06:11:36.053682 138534581335872 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:11:36.084128 138534581335872 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 06:11:36.084198 138534581335872 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 06:11:36.104260 138534581335872 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 06:11:36.104323 138534581335872 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 06:11:47.166786 138534581335872 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0417 06:11:47.176642 138534581335872 metric_logger.py:289] number parameters: 1.104 billion I0417 06:11:58.753469 138534581335872 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 06:11:58.905282 138534581335872 checkpointing.py:798] Waited 0.15179443359375 seconds for step 0 to finish before starting checkpointing. I0417 06:11:58.907711 138534581335872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:11:58.909380 138534581335872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 06:11:58.911086 138534581335872 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0. I0417 06:11:59.641014 138534581335872 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 06:11:59.660465 138534581335872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:11:59.660533 138534581335872 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:11:59.895144 138405446936320 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0 I0417 06:11:59.938523 138534581335872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.279480s I0417 06:11:59.939378 138534581335872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.311 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2904539108276367 s) (per-host) I0417 06:11:59.939448 138534581335872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290544s (batch_requests_ready=0.003055s, total_serialization_initiated=0.286718s, others=0.000771s) I0417 06:11:59.939530 138534581335872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294471s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.294454) I0417 06:11:59.940355 138534581335872 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0. I0417 06:11:59.940638 138405455329024 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:31:59.940605 I0417 06:11:59.946519 138534581335872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:11:59.946763 138404338808576 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:11:59.946901 138534581335872 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406318.9076936, 'wait_for_prev_duration_secs': 5.793571472167969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406318.9094167, 'checkpointer_blocking_duration_secs': 1.0313596725463867, 'get_old_steps_start_time': 1776406319.9408002, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776406318.905664, 'checkpoint_manager_blocking_duration_secs': 1.0411980152130127} I0417 06:11:59.947008 138534581335872 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 06:11:59.947077 138534581335872 max_utils.py:750] Memstats: After params initialized: I0417 06:11:59.947131 138534581335872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 06:11:59.947164 138534581335872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 06:11:59.947190 138534581335872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 06:11:59.947214 138534581335872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 06:12:00.287563 138534581335872 metric_logger.py:185] completed step: 0, seconds: 11.287, TFLOP/s/device: 1.204, Tokens/s/device: 181.455, total_weights: 65536, loss: 10.874 I0417 06:12:00.290410 138534581335872 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_15_ocdbt_false/tensorboard/' I0417 06:12:00.762925 138534581335872 metric_logger.py:185] completed step: 1, seconds: 1.531, TFLOP/s/device: 8.877, Tokens/s/device: 1337.996, total_weights: 65536, loss: 10.874 I0417 06:12:00.904793 138534581335872 metric_logger.py:185] completed step: 2, seconds: 0.476, TFLOP/s/device: 28.521, Tokens/s/device: 4298.927, total_weights: 65536, loss: 10.560 I0417 06:12:01.052578 138534581335872 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1140.530, Tokens/s/device: 171913.036, total_weights: 65536, loss: 9.979 I0417 06:12:01.064799 138534581335872 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0417 06:12:01.347542 138534581335872 checkpointing.py:798] Waited 0.2827115058898926 seconds for step 5 to finish before starting checkpointing. I0417 06:12:01.350312 138534581335872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:12:02.119861 138405430150912 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/items I0417 06:12:02.267448 138402714109696 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776406321416820466, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776406321416820466, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0417 06:12:03.273569 2810 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 06:12:05.024009 138404862793472 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_0 I0417 06:12:13.869249 138404381587200 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.606899s (commit=9.555287s, array_metadata_write=1.051612s) I0417 06:12:13.870599 138405455329024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 111.070 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 14.221633434295654 s) (per-host) I0417 06:12:13.870774 138405455329024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 13.930026s. I0417 06:12:14.514441 138405455329024 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776406321416820466, '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_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0417 06:12:15.240489 138402714109696 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776406321416820466, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA W0417 06:12:15.283385 138405455329024 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/items, I0417 06:12:15.699025 138405455329024 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000813s. I0417 06:12:15.699330 138405455329024 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.066720s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/items I0417 06:12:15.701326 138405455329024 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0/items I0417 06:12:16.360526 138405455329024 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0 I0417 06:12:17.720530 138405455329024 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0`. I0417 06:12:17.722903 138405455329024 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.81 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/0 I0417 06:12:17.724472 138405455329024 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.783720s. I0417 06:12:17.724730 138404338808576 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:12:17.724911 138404338808576 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:12:17.725016 138404338808576 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:12:17.727294 138404338808576 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 06:12:17.727596 138534581335872 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 06:12:17.727763 138534581335872 checkpoint_manager.py:1452] Waiting for previous save to complete took 16.377470 seconds. If this number is high, consider checkpointing less frequently. I0417 06:12:17.729560 138534581335872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0417 06:12:17.731932 138534581335872 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5. I0417 06:12:18.087865 138534581335872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:12:18.088003 138534581335872 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:12:18.126555 138534581335872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041749s I0417 06:12:18.127516 138534581335872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 25.203 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06120586395263672 s) (per-host) I0417 06:12:18.127624 138534581335872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.061337s (batch_requests_ready=0.006264s, total_serialization_initiated=0.054215s, others=0.000858s) I0417 06:12:18.127755 138534581335872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.066435s (all_items=0.000032s, per_item={'items': '0.00003242'}, temp_paths=0.066402) I0417 06:12:18.128708 138534581335872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.40 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5. I0417 06:12:18.129156 138404338808576 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:32:18.129115 I0417 06:12:18.131625 138534581335872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:12:18.131942 138403844716288 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:12:18.132123 138534581335872 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406321.3502634, 'wait_for_prev_duration_secs': 16.37747025489807, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776406337.7296293, 'checkpointer_blocking_duration_secs': 0.3997151851654053, 'get_old_steps_start_time': 1776406338.129372, 'get_old_steps_duration_secs': 4.029273986816406e-05, 'checkpoint_manager_blocking_start_time': 1776406321.3480225, 'checkpoint_manager_blocking_duration_secs': 16.784050464630127} I0417 06:12:18.132270 138534581335872 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0417 06:12:18.313569 138405455329024 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5 I0417 06:12:20.089270 138404884887296 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5/items I0417 06:12:23.277529 138404862793472 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_0 I0417 06:12:31.242252 138404381587200 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 9.728114s (commit=8.574320s, array_metadata_write=1.153794s) I0417 06:12:31.243382 138404338808576 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 119.875 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 13.177055597305298 s) (per-host) I0417 06:12:31.243513 138404338808576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 13.114210s. W0417 06:12:32.651751 138404338808576 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5/items, I0417 06:12:33.579356 138404338808576 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000409s. I0417 06:12:33.579581 138404338808576 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.558270s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5/items I0417 06:12:33.581112 138404338808576 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5/items I0417 06:12:34.137913 138404338808576 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5 I0417 06:12:35.755386 138404338808576 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5`. I0417 06:12:35.756202 138404338808576 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.03 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/5 I0417 06:12:35.757530 138404338808576 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.628225s. I0417 06:12:35.757672 138403844716288 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:12:35.757747 138403844716288 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:12:35.757821 138403844716288 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:12:35.759555 138403844716288 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 06:12:41.105784 138534581335872 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.766, Tokens/s/device: 14284.120, total_weights: 65536, loss: 9.457 I0417 06:12:41.120115 138534581335872 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.204, Tokens/s/device: 13898.031, total_weights: 65536, loss: 8.955 I0417 06:12:41.260587 138534581335872 metric_logger.py:185] completed step: 6, seconds: 40.055, TFLOP/s/device: 0.339, Tokens/s/device: 51.130, total_weights: 65536, loss: 8.465 I0417 06:12:41.408313 138534581335872 metric_logger.py:185] completed step: 7, seconds: 0.009, TFLOP/s/device: 1585.984, Tokens/s/device: 239056.846, total_weights: 65536, loss: 7.998 I0417 06:12:41.556079 138534581335872 metric_logger.py:185] completed step: 8, seconds: 0.142, TFLOP/s/device: 95.555, Tokens/s/device: 14403.061, total_weights: 65536, loss: 7.567 I0417 06:12:41.704378 138534581335872 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 06:12:41.705093 138534581335872 checkpointing.py:798] Waited 0.0007293224334716797 seconds for step 9 to finish before starting checkpointing. I0417 06:12:41.707155 138534581335872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:12:41.708586 138534581335872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 06:12:41.710008 138534581335872 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9. I0417 06:12:42.420064 138534581335872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 06:12:42.420165 138534581335872 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 06:12:42.447341 138534581335872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028709s I0417 06:12:42.448174 138534581335872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.523 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037149667739868164 s) (per-host) I0417 06:12:42.448248 138534581335872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037247s (batch_requests_ready=0.002537s, total_serialization_initiated=0.033947s, others=0.000763s) I0417 06:12:42.448329 138534581335872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041152s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.041135) I0417 06:12:42.449047 138534581335872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9. I0417 06:12:42.449391 138403844716288 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 06:32:42.449356 I0417 06:12:42.456310 138534581335872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 06:12:42.456550 138404330415872 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 06:12:42.456680 138534581335872 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776406361.7071228, 'wait_for_prev_duration_secs': 7.295608520507812e-05, 'time_between_consecutive_saves_sec': 5.947526931762695, 'checkpointer_blocking_start_time': 1776406361.7086253, 'checkpointer_blocking_duration_secs': 0.7408969402313232, 'get_old_steps_start_time': 1776406362.4495418, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776406361.705337, 'checkpoint_manager_blocking_duration_secs': 0.7513110637664795} I0417 06:12:42.456798 138534581335872 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 06:12:42.456857 138534581335872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 06:12:42.654808 138404338808576 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9 I0417 06:12:44.812448 138405455329024 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/items I0417 06:12:47.868433 138404862793472 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_0 I0417 06:12:55.551307 138404381587200 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 9.647662s (commit=8.556970s, array_metadata_write=1.090692s) I0417 06:12:55.552472 138403844716288 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 120.200 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 13.141422033309937 s) (per-host) I0417 06:12:55.552594 138403844716288 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 13.103096s. W0417 06:12:56.966249 138403844716288 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/items, I0417 06:12:57.388455 138403844716288 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000452s. I0417 06:12:57.388701 138403844716288 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.058679s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/items I0417 06:12:57.390149 138403844716288 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9/items I0417 06:12:57.913289 138403844716288 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9 I0417 06:12:59.721078 138403844716288 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9`. I0417 06:12:59.721734 138403844716288 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.01 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/nnx_xpk_feat_nnx_post_train_fixes_20260417_023710_15_ocdbt_false/checkpoints/9 I0417 06:12:59.723042 138403844716288 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 17.273540s. I0417 06:12:59.723253 138404330415872 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 06:12:59.723376 138404330415872 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 06:12:59.723430 138404330415872 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 06:12:59.725924 138404330415872 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 06:12:59.726110 138534581335872 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 06:12:59.726289 138534581335872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 06:12:59.727016 138534581335872 metric_logger.py:185] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.742, Tokens/s/device: 13828.401, 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 06:13:12 UTC 2026 EXIT_CODE=0