XPK Start: Thu Apr 16 22:41:37 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-16 22:42:01.351165: 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) I0416 22:42:01.878133 132042903070528 max_utils.py:273] Attempting to initialize the jax distributed system... I0416 22:42:10.918578 132042903070528 distributed.py:149] Starting JAX distributed service on [::]:8482 I0416 22:42:10.921142 132042903070528 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-lfdjq-slice-job-0-0.mt-09-pdb-lt-1-lfdjq:8482 I0416 22:42:12.081540 132042903070528 max_utils.py:284] Jax distributed system initialized! I0416 22:42:18.200937 132042903070528 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 22:42:18.201036 132042903070528 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 22:42:18.201077 132042903070528 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 I0416 22:42:18.201112 132042903070528 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 22:42:18.863013 132042903070528 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0416 22:42:18.975115 132042903070528 checkpointing.py:688] Setting up checkpoint logger... I0416 22:42:18.975227 132042903070528 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 22:42:18.975275 132042903070528 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0416 22:42:18.975473 132042903070528 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 0x7816ea948500>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 22:42:21.813596 132042903070528 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 22:42:21.813828 132042903070528 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 0x7802c448f7a0>}, handler_registry=None I0416 22:42:21.814087 132042903070528 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7802c448f7a0>` 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`. I0416 22:42:21.814138 132042903070528 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7802c4457380>` 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`. I0416 22:42:21.814174 132042903070528 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 0x7802c448f7a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7802c448f7a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7802c4457380>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7802c4457380>}). I0416 22:42:21.814489 132042903070528 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0416 22:42:21.814558 132042903070528 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7802e4115ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0416 22:42:22.943509 132042903070528 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints I0416 22:42:23.361033 132042903070528 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7802c44ba360> I0416 22:42:23.361207 132042903070528 checkpointing.py:302] Checkpoint manager created! I0416 22:42:24.300273 132042903070528 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 22:42:24.300391 132042903070528 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0416 22:42:25.683439 132042903070528 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: ('tensor',) I0416 22:42:25.683556 132042903070528 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0416 22:42:25.683610 132042903070528 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0416 22:42:25.683653 132042903070528 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: ('tensor', None, 'fsdp') I0416 22:42:25.683690 132042903070528 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0416 22:42:25.683726 132042903070528 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0416 22:42:25.683761 132042903070528 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0416 22:42:25.683798 132042903070528 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: ('tensor', None, None, 'fsdp') I0416 22:42:25.683851 132042903070528 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0416 22:42:25.683889 132042903070528 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0416 22:42:25.683924 132042903070528 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', 'tensor') I0416 22:42:25.683956 132042903070528 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: ('tensor', 'fsdp') I0416 22:42:26.081220 132042903070528 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:42:26.081314 132042903070528 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0416 22:42:26.086541 132042903070528 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 22:42:26.086596 132042903070528 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0416 22:42:26.102708 132042903070528 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:42:26.102764 132042903070528 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0416 22:42:26.118140 132042903070528 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 22:42:26.118199 132042903070528 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0416 22:42:26.141959 132042903070528 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:42:26.142027 132042903070528 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0416 22:42:26.157662 132042903070528 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:42:26.157723 132042903070528 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0416 22:42:26.173226 132042903070528 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 22:42:26.173284 132042903070528 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0416 22:42:26.202445 132042903070528 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 22:42:26.202517 132042903070528 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0416 22:42:26.222204 132042903070528 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 22:42:26.222273 132042903070528 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0416 22:42:41.537027 132042903070528 max_utils.py:791] Total memory size: 0.9 GB, Output size: 0.4 GB, Temp size: 0.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0416 22:42:41.546030 132042903070528 metric_logger.py:289] number parameters: 1.104 billion I0416 22:42:42.765983 132042903070528 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 22:42:56.513319 132042903070528 checkpointing.py:798] Waited 13.74731969833374 seconds for step 0 to finish before starting checkpointing. I0416 22:42:56.515524 132042903070528 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:42:56.517629 132042903070528 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0416 22:42:56.519107 132042903070528 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0. I0416 22:42:57.249763 132042903070528 signaling_client.py:364] Using JaxDistributedSignalingClient I0416 22:42:57.269527 132042903070528 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:42:57.269599 132042903070528 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 I0416 22:42:57.515311 131914534999808 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0 I0416 22:42:57.532866 132042903070528 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.264733s I0416 22:42:57.533642 132042903070528 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.595 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27571654319763184 s) (per-host) I0416 22:42:57.533720 132042903070528 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.275815s (batch_requests_ready=0.003256s, total_serialization_initiated=0.271864s, others=0.000694s) I0416 22:42:57.533807 132042903070528 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.279911s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.279894) I0416 22:42:57.534650 132042903070528 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0. I0416 22:42:57.535027 131914543392512 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 23:02:57.534989 I0416 22:42:57.542171 132042903070528 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:42:57.542477 131914014914304 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:42:57.542632 132042903070528 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776379376.515507, 'wait_for_prev_duration_secs': 5.888938903808594e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776379376.517667, 'checkpointer_blocking_duration_secs': 1.017507791519165, 'get_old_steps_start_time': 1776379377.535202, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776379376.5137053, 'checkpoint_manager_blocking_duration_secs': 1.0288872718811035} I0416 22:42:57.542735 132042903070528 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 22:42:57.542798 132042903070528 max_utils.py:750] Memstats: After params initialized: I0416 22:42:57.542863 132042903070528 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0416 22:42:57.542899 132042903070528 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0416 22:42:57.542926 132042903070528 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0416 22:42:57.542950 132042903070528 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0416 22:42:57.897045 132042903070528 metric_logger.py:185] completed step: 0, seconds: 0.889, TFLOP/s/device: 3.820, Tokens/s/device: 575.846, total_weights: 16384, loss: 10.864 I0416 22:42:57.900131 132042903070528 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/tensorboard/' I0416 22:42:58.359887 132042903070528 metric_logger.py:185] completed step: 1, seconds: 15.128, TFLOP/s/device: 0.225, Tokens/s/device: 33.845, total_weights: 16384, loss: 10.864 I0416 22:42:58.412634 132042903070528 metric_logger.py:185] completed step: 2, seconds: 0.464, TFLOP/s/device: 7.324, Tokens/s/device: 1103.919, total_weights: 16384, loss: 9.760 I0416 22:42:58.471333 132042903070528 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 280.147, Tokens/s/device: 42226.804, total_weights: 16384, loss: 8.820 I0416 22:42:59.292755 131914518214400 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/items I0416 22:42:59.461759 131913996166912 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776379379050952959, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776379379050952959, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0416 22:43:00.818533 2555 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0416 22:43:03.696346 131918306825984 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_0 I0416 22:43:13.555858 132042903070528 metric_logger.py:185] completed step: 4, seconds: 0.054, TFLOP/s/device: 63.000, Tokens/s/device: 9496.077, total_weights: 16384, loss: 7.947 I0416 22:43:13.566278 132042903070528 metric_logger.py:185] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.446, Tokens/s/device: 8809.663, total_weights: 16384, loss: 7.209 I0416 22:43:13.621783 132042903070528 metric_logger.py:185] completed step: 6, seconds: 15.084, TFLOP/s/device: 0.225, Tokens/s/device: 33.944, total_weights: 16384, loss: 6.651 I0416 22:43:13.680389 132042903070528 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 449.369, Tokens/s/device: 67733.827, total_weights: 16384, loss: 6.267 I0416 22:43:13.738980 132042903070528 metric_logger.py:185] completed step: 8, seconds: 0.057, TFLOP/s/device: 59.934, Tokens/s/device: 9033.966, total_weights: 16384, loss: 6.030 I0416 22:43:13.798418 132042903070528 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 22:43:13.799086 132042903070528 checkpointing.py:798] Waited 0.0006775856018066406 seconds for step 9 to finish before starting checkpointing. I0416 22:43:13.801049 132042903070528 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:43:36.229187 131914501428992 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.417169s (commit=34.234846s, array_metadata_write=1.182323s) I0416 22:43:36.230316 131914543392512 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 40.531 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.97236371040344 s) (per-host) I0416 22:43:36.230436 131914543392512 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.695301s. I0416 22:43:36.974195 131914543392512 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776379379050952959, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0416 22:43:38.566977 131914543392512 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000447s. I0416 22:43:38.758334 131914543392512 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:43:39.399415 131913996166912 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776379379050952959, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0416 22:43:39.693877 131914543392512 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.174718s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/items I0416 22:43:39.695531 131914543392512 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0/items I0416 22:43:40.664222 131914543392512 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0 I0416 22:43:42.055299 131914543392512 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0`. I0416 22:43:42.056067 131914543392512 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 45.54 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/0 I0416 22:43:42.057406 131914543392512 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 44.522269s. I0416 22:43:42.057631 131914014914304 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:43:42.057752 131914014914304 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:43:42.057815 131914014914304 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:43:42.059721 131914014914304 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 22:43:42.059915 132042903070528 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. W0416 22:43:42.060048 132042903070528 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.258996 seconds. If this number is high, consider checkpointing less frequently. I0416 22:43:42.061938 132042903070528 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0416 22:43:42.064065 132042903070528 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9. I0416 22:43:42.763593 132042903070528 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0416 22:43:42.763685 132042903070528 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 I0416 22:43:42.791722 132042903070528 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029518s I0416 22:43:42.792481 132042903070528 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.292 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0382847785949707 s) (per-host) I0416 22:43:42.792547 132042903070528 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038367s (batch_requests_ready=0.002849s, total_serialization_initiated=0.034832s, others=0.000686s) I0416 22:43:42.792623 132042903070528 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042288s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.042272) I0416 22:43:42.793322 132042903070528 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9. I0416 22:43:42.793651 131914014914304 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-16 23:03:42.793612 I0416 22:43:42.795697 132042903070528 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 22:43:42.795978 131912806954752 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 22:43:42.796118 132042903070528 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776379393.8010192, 'wait_for_prev_duration_secs': 28.25899600982666, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776379422.061977, 'checkpointer_blocking_duration_secs': 0.7318129539489746, 'get_old_steps_start_time': 1776379422.79382, 'get_old_steps_duration_secs': 4.5299530029296875e-05, 'checkpoint_manager_blocking_start_time': 1776379393.799326, 'checkpoint_manager_blocking_duration_secs': 28.996755838394165} I0416 22:43:42.796297 132042903070528 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 22:43:42.796343 132042903070528 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 22:43:42.992027 131908646205184 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9 I0416 22:43:45.170163 131912883619584 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9/items I0416 22:43:48.005553 131913478043392 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_0 I0416 22:44:27.335738 131913444472576 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.070241s (commit=38.605820s, array_metadata_write=2.464421s) I0416 22:44:27.336977 131914014914304 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.431 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.58272409439087 s) (per-host) I0416 22:44:27.337122 131914014914304 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.543354s. I0416 22:44:29.532413 131914014914304 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000403s. I0416 22:44:29.788537 131914014914304 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0416 22:44:30.724037 131914014914304 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.169875s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9/items I0416 22:44:30.725606 131914014914304 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9/items I0416 22:44:32.082293 131914014914304 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9 I0416 22:44:33.504222 131914014914304 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_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9`. I0416 22:44:33.504937 131914014914304 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.44 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260416_221011/nnx_xpk_feat_nnx_post_train_fixes_20260416_221011_09_pdb_lt_1/checkpoints/9 I0416 22:44:33.506957 131914014914304 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.713189s. I0416 22:44:33.507157 131912806954752 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 22:44:33.507272 131912806954752 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 22:44:33.507333 131912806954752 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 22:44:33.511946 131912806954752 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 22:44:33.512088 132042903070528 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. I0416 22:44:33.512207 132042903070528 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 22:44:33.512868 132042903070528 metric_logger.py:185] completed step: 9, seconds: 0.059, TFLOP/s/device: 58.012, Tokens/s/device: 8744.215, total_weights: 16384, loss: 5.890 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Thu Apr 16 22:44:47 UTC 2026 EXIT_CODE=0