XPK Start: Fri Apr 17 00:34:51 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 00:35:15.850978: 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 00:35:16.026936 137142910457664 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 00:35:25.067487 137142910457664 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 00:35:25.070076 137142910457664 distributed.py:172] Connecting to JAX distributed service on mt-01-base-mqjk4-slice-job-0-0.mt-01-base-mqjk4:8482 I0417 00:35:26.248519 137142910457664 max_utils.py:284] Jax distributed system initialized! I0417 00:35:32.626592 137142910457664 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 00:35:32.626694 137142910457664 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 00:35:32.626736 137142910457664 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 00:35:32.626772 137142910457664 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 00:35:33.301577 137142910457664 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 00:35:33.416088 137142910457664 checkpointing.py:688] Setting up checkpoint logger... I0417 00:35:33.416197 137142910457664 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 00:35:33.416241 137142910457664 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 00:35:33.416438 137142910457664 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 0x7cba7971c410>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 00:35:37.810538 137142910457664 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 00:35:37.810764 137142910457664 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 0x7ca5e46b56d0>}, handler_registry=None I0417 00:35:37.811024 137142910457664 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ca5e46b56d0>` 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 00:35:37.811075 137142910457664 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ca4fc42ddf0>` 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 00:35:37.811111 137142910457664 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 0x7ca5e46b56d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ca5e46b56d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ca4fc42ddf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ca4fc42ddf0>}). I0417 00:35:37.811442 137142910457664 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 00:35:37.811512 137142910457664 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ca4fc40e020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 00:35:38.954401 137142910457664 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_01_base/checkpoints I0417 00:35:39.386026 137142910457664 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ca5e42676e0> I0417 00:35:39.386201 137142910457664 checkpointing.py:302] Checkpoint manager created! I0417 00:35:40.327490 137142910457664 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 00:35:40.327611 137142910457664 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 00:35:41.702270 137142910457664 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 00:35:41.702386 137142910457664 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 00:35:41.702439 137142910457664 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 00:35:41.702481 137142910457664 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 00:35:41.702538 137142910457664 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 00:35:41.702576 137142910457664 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 00:35:41.702613 137142910457664 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:35:41.702647 137142910457664 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 00:35:41.702681 137142910457664 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:35:41.702712 137142910457664 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 00:35:41.702744 137142910457664 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 00:35:41.702777 137142910457664 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 00:35:42.101189 137142910457664 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 00:35:42.101280 137142910457664 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 00:35:42.106536 137142910457664 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 00:35:42.106590 137142910457664 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 00:35:42.122660 137142910457664 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 00:35:42.122725 137142910457664 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 00:35:42.138018 137142910457664 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 00:35:42.138073 137142910457664 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 00:35:42.161676 137142910457664 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 00:35:42.161744 137142910457664 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 00:35:42.177317 137142910457664 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 00:35:42.177377 137142910457664 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 00:35:42.192806 137142910457664 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 00:35:42.192884 137142910457664 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 00:35:42.222461 137142910457664 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 00:35:42.222530 137142910457664 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 00:35:42.242114 137142910457664 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 00:35:42.242183 137142910457664 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 00:35:52.781021 137142910457664 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 00:35:52.789993 137142910457664 metric_logger.py:289] number parameters: 1.104 billion I0417 00:36:04.281161 137142910457664 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 00:36:04.431986 137142910457664 checkpointing.py:798] Waited 0.15081477165222168 seconds for step 0 to finish before starting checkpointing. I0417 00:36:04.434677 137142910457664 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 00:36:04.436292 137142910457664 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 00:36:04.437742 137142910457664 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_01_base/checkpoints/0. I0417 00:36:05.149346 137142910457664 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 00:36:05.171000 137142910457664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 00:36:05.171071 137142910457664 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 00:36:05.419364 137013402609408 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_01_base/checkpoints/0 I0417 00:36:05.448018 137142910457664 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278437s I0417 00:36:05.450136 137142910457664 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.304 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2908186912536621 s) (per-host) I0417 00:36:05.450207 137142910457664 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290907s (batch_requests_ready=0.003300s, total_serialization_initiated=0.285570s, others=0.002038s) I0417 00:36:05.450291 137142910457664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294941s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.294922) I0417 00:36:05.451115 137142910457664 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 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_01_base/checkpoints/0. I0417 00:36:05.451451 137013411002112 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 00:56:05.451413 I0417 00:36:05.456352 137142910457664 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 00:36:05.456649 137011230922496 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 00:36:05.456798 137142910457664 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776386164.4346588, 'wait_for_prev_duration_secs': 6.0558319091796875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776386164.4363306, 'checkpointer_blocking_duration_secs': 1.0152721405029297, 'get_old_steps_start_time': 1776386165.4516282, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776386164.43238, 'checkpoint_manager_blocking_duration_secs': 1.0243809223175049} I0417 00:36:05.456922 137142910457664 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 00:36:05.456992 137142910457664 max_utils.py:750] Memstats: After params initialized: I0417 00:36:05.457059 137142910457664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 00:36:05.457101 137142910457664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 00:36:05.457129 137142910457664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 00:36:05.457153 137142910457664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 00:36:05.800454 137142910457664 metric_logger.py:185] completed step: 0, seconds: 11.209, TFLOP/s/device: 1.212, Tokens/s/device: 182.714, total_weights: 65536, loss: 10.874 I0417 00:36:05.803793 137142910457664 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_01_base/tensorboard/' I0417 00:36:06.234519 137142910457664 metric_logger.py:185] completed step: 1, seconds: 1.516, TFLOP/s/device: 8.963, Tokens/s/device: 1351.019, total_weights: 65536, loss: 10.874 I0417 00:36:06.376307 137142910457664 metric_logger.py:185] completed step: 2, seconds: 0.435, TFLOP/s/device: 31.236, Tokens/s/device: 4708.252, total_weights: 65536, loss: 10.262 I0417 00:36:06.523733 137142910457664 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1113.882, Tokens/s/device: 167896.376, total_weights: 65536, loss: 9.732 I0417 00:36:07.700576 137013385824000 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_01_base/checkpoints/0/items I0417 00:36:07.886937 137010677999360 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386167449962654, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776386167449962654, "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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 00:36:09.261890 2864 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 00:36:11.922172 137012824028928 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_01_base/checkpoints/0/items/array_metadatas/process_0 I0417 00:36:30.645502 137142910457664 metric_logger.py:185] completed step: 4, seconds: 0.144, TFLOP/s/device: 94.525, Tokens/s/device: 14247.849, total_weights: 65536, loss: 9.274 I0417 00:36:30.656192 137142910457664 metric_logger.py:185] completed step: 5, seconds: 0.146, TFLOP/s/device: 93.068, Tokens/s/device: 14028.166, total_weights: 65536, loss: 8.888 I0417 00:36:30.800655 137142910457664 metric_logger.py:185] completed step: 6, seconds: 24.121, TFLOP/s/device: 0.563, Tokens/s/device: 84.905, total_weights: 65536, loss: 8.589 I0417 00:36:30.948276 137142910457664 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1601.878, Tokens/s/device: 241452.488, total_weights: 65536, loss: 8.380 I0417 00:36:31.095894 137142910457664 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.148, Tokens/s/device: 14040.284, total_weights: 65536, loss: 8.251 I0417 00:36:31.244380 137142910457664 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 00:36:31.245074 137142910457664 checkpointing.py:798] Waited 0.0007073879241943359 seconds for step 9 to finish before starting checkpointing. I0417 00:36:31.246990 137142910457664 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 00:36:45.591016 137011239315200 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.341456s (commit=35.101793s, array_metadata_write=1.239664s) I0417 00:36:45.593047 137013411002112 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.066 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.433701038360596 s) (per-host) I0417 00:36:45.593172 137013411002112 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.141608s. I0417 00:36:47.211727 137013411002112 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386167449962654, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 00:36:48.246449 137010677999360 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776386167449962654, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0417 00:36:48.325263 137013411002112 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000458s. I0417 00:36:48.715886 137013411002112 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 00:36:49.673437 137013411002112 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.334661s. 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_01_base/checkpoints/0/items I0417 00:36:49.675237 137013411002112 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_01_base/checkpoints/0/items I0417 00:36:50.380133 137013411002112 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_01_base/checkpoints/0 I0417 00:36:52.740797 137013411002112 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_01_base/checkpoints/0`. I0417 00:36:52.741553 137013411002112 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 48.31 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_01_base/checkpoints/0 I0417 00:36:52.743037 137013411002112 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 47.291471s. I0417 00:36:52.743173 137011230922496 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 00:36:52.743238 137011230922496 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 00:36:52.743301 137011230922496 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 00:36:52.745373 137011230922496 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 00:36:52.745536 137142910457664 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 00:36:52.745677 137142910457664 checkpoint_manager.py:1452] Waiting for previous save to complete took 21.498686 seconds. If this number is high, consider checkpointing less frequently. I0417 00:36:52.747467 137142910457664 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 00:36:52.749451 137142910457664 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_01_base/checkpoints/9. I0417 00:36:53.044057 137142910457664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 00:36:53.044151 137142910457664 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 00:36:53.071481 137142910457664 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 00:36:53.072367 137142910457664 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.534 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.038056373596191406 s) (per-host) I0417 00:36:53.072451 137142910457664 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038162s (batch_requests_ready=0.002889s, total_serialization_initiated=0.034467s, others=0.000806s) I0417 00:36:53.072546 137142910457664 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042148s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.042130) I0417 00:36:53.073340 137142910457664 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.33 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_01_base/checkpoints/9. I0417 00:36:53.073642 137011230922496 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 00:56:53.073608 I0417 00:36:53.085361 137142910457664 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 00:36:53.085655 137011797157632 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 00:36:53.085790 137142910457664 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776386191.2469578, 'wait_for_prev_duration_secs': 21.49868607521057, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776386212.7475057, 'checkpointer_blocking_duration_secs': 0.32627153396606445, 'get_old_steps_start_time': 1776386213.0738006, 'get_old_steps_duration_secs': 3.24249267578125e-05, 'checkpoint_manager_blocking_start_time': 1776386191.2453122, 'checkpoint_manager_blocking_duration_secs': 21.840441703796387} I0417 00:36:53.085910 137142910457664 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 00:36:53.085955 137142910457664 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 00:36:53.273202 137013411002112 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_01_base/checkpoints/9 I0417 00:36:55.883010 137012300416768 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_01_base/checkpoints/9/items I0417 00:36:59.153338 137012292024064 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_01_base/checkpoints/9/items/array_metadatas/process_0 I0417 00:37:38.980590 137011805550336 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.573147s (commit=40.178157s, array_metadata_write=1.394990s) I0417 00:37:38.981780 137011230922496 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.378 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.947447538375854 s) (per-host) I0417 00:37:38.981935 137011230922496 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.908187s. I0417 00:37:41.120890 137011230922496 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 00:37:41.466989 137011230922496 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000426s. I0417 00:37:42.182467 137011230922496 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.893192s. 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_01_base/checkpoints/9/items I0417 00:37:42.184306 137011230922496 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_01_base/checkpoints/9/items I0417 00:37:42.786674 137011230922496 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_01_base/checkpoints/9 I0417 00:37:44.566904 137011230922496 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_01_base/checkpoints/9`. I0417 00:37:44.567727 137011230922496 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.82 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_01_base/checkpoints/9 I0417 00:37:44.569281 137011230922496 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.495530s. I0417 00:37:44.569494 137011797157632 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 00:37:44.569611 137011797157632 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 00:37:44.569674 137011797157632 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 00:37:44.571699 137011797157632 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 00:37:44.571887 137142910457664 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 00:37:44.572064 137142910457664 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 00:37:44.572772 137142910457664 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.330, Tokens/s/device: 13917.014, 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 00:37:52 UTC 2026 EXIT_CODE=0