XPK Start: Fri Apr 17 01:14:13 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 01:14:37.015336: 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 01:14:37.191726 133173294679872 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 01:14:46.232573 133173294679872 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 01:14:46.234938 133173294679872 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-xqr5k-slice-job-0-0.mt-09-pdb-lt-1-xqr5k:8482 I0417 01:14:47.245130 133173294679872 max_utils.py:284] Jax distributed system initialized! I0417 01:14:53.531263 133173294679872 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 01:14:53.531368 133173294679872 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 01:14:53.531408 133173294679872 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 01:14:53.531442 133173294679872 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 01:14:54.202327 133173294679872 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0417 01:14:54.315011 133173294679872 checkpointing.py:688] Setting up checkpoint logger... I0417 01:14:54.315123 133173294679872 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 01:14:54.315171 133173294679872 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 01:14:54.315365 133173294679872 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 0x791e3ef18170>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 01:14:57.157347 133173294679872 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 01:14:57.157573 133173294679872 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 0x7909dc3338f0>}, handler_registry=None I0417 01:14:57.157807 133173294679872 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7909dc3338f0>` 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 01:14:57.157867 133173294679872 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7909f419f260>` 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 01:14:57.157904 133173294679872 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 0x7909dc3338f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7909dc3338f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7909f419f260>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7909f419f260>}). I0417 01:14:57.158252 133173294679872 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 01:14:57.158324 133173294679872 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7909dc78e020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 01:14:58.724150 133173294679872 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_09_pdb_lt_1/checkpoints I0417 01:14:58.735420 133173294679872 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_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7909dc388bf0> I0417 01:14:58.735536 133173294679872 checkpointing.py:302] Checkpoint manager created! I0417 01:14:59.687937 133173294679872 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 01:14:59.688055 133173294679872 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0417 01:15:01.135102 133173294679872 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: ('tensor',) I0417 01:15:01.135220 133173294679872 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0417 01:15:01.135275 133173294679872 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0417 01:15:01.135319 133173294679872 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: ('tensor', None, 'fsdp') I0417 01:15:01.135360 133173294679872 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0417 01:15:01.135398 133173294679872 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0417 01:15:01.135435 133173294679872 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0417 01:15:01.135470 133173294679872 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: ('tensor', None, None, 'fsdp') I0417 01:15:01.135504 133173294679872 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0417 01:15:01.135536 133173294679872 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0417 01:15:01.135569 133173294679872 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', 'tensor') I0417 01:15:01.135602 133173294679872 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: ('tensor', 'fsdp') I0417 01:15:01.530731 133173294679872 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 01:15:01.530821 133173294679872 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 01:15:01.536145 133173294679872 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 01:15:01.536199 133173294679872 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 01:15:01.552382 133173294679872 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 01:15:01.552445 133173294679872 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 01:15:01.568012 133173294679872 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 01:15:01.568076 133173294679872 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0417 01:15:01.591962 133173294679872 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 01:15:01.592025 133173294679872 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 01:15:01.607645 133173294679872 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 01:15:01.607703 133173294679872 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 01:15:01.623274 133173294679872 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 01:15:01.623332 133173294679872 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 01:15:01.652817 133173294679872 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 01:15:01.652918 133173294679872 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0417 01:15:01.672627 133173294679872 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 01:15:01.672695 133173294679872 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0417 01:15:16.682242 133173294679872 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. I0417 01:15:16.691089 133173294679872 metric_logger.py:289] number parameters: 1.104 billion I0417 01:15:17.919774 133173294679872 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 01:15:31.893149 133173294679872 checkpointing.py:798] Waited 13.973350763320923 seconds for step 0 to finish before starting checkpointing. I0417 01:15:31.895442 133173294679872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 01:15:31.897195 133173294679872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 01:15:31.898580 133173294679872 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_09_pdb_lt_1/checkpoints/0. I0417 01:15:32.610578 133173294679872 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 01:15:32.629963 133173294679872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 01:15:32.630031 133173294679872 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 01:15:32.874381 133045953820416 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_09_pdb_lt_1/checkpoints/0 I0417 01:15:32.889363 133173294679872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.260803s I0417 01:15:32.890205 133173294679872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.675 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27184104919433594 s) (per-host) I0417 01:15:32.890285 133173294679872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.271942s (batch_requests_ready=0.003325s, total_serialization_initiated=0.267847s, others=0.000769s) I0417 01:15:32.890367 133173294679872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.275899s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.275882) I0417 01:15:32.891193 133173294679872 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.99 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_09_pdb_lt_1/checkpoints/0. I0417 01:15:32.891477 133045962213120 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:35:32.891445 I0417 01:15:32.905100 133173294679872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 01:15:32.905373 133044908312320 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 01:15:32.905531 133173294679872 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_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776388531.895422, 'wait_for_prev_duration_secs': 7.152557373046875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776388531.89724, 'checkpointer_blocking_duration_secs': 0.9943804740905762, 'get_old_steps_start_time': 1776388532.8916483, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776388531.8935819, 'checkpoint_manager_blocking_duration_secs': 1.0119106769561768} I0417 01:15:32.905636 133173294679872 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 01:15:32.905703 133173294679872 max_utils.py:750] Memstats: After params initialized: I0417 01:15:32.905751 133173294679872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 01:15:32.905782 133173294679872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 01:15:32.905808 133173294679872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 01:15:32.905831 133173294679872 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 01:15:33.250310 133173294679872 metric_logger.py:185] completed step: 0, seconds: 0.902, TFLOP/s/device: 3.766, Tokens/s/device: 567.657, total_weights: 16384, loss: 10.864 I0417 01:15:33.253370 133173294679872 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_09_pdb_lt_1/tensorboard/' I0417 01:15:33.676141 133173294679872 metric_logger.py:185] completed step: 1, seconds: 15.327, TFLOP/s/device: 0.222, Tokens/s/device: 33.405, total_weights: 16384, loss: 10.864 I0417 01:15:33.728833 133173294679872 metric_logger.py:185] completed step: 2, seconds: 0.427, TFLOP/s/device: 7.960, Tokens/s/device: 1199.844, total_weights: 16384, loss: 9.760 I0417 01:15:33.787405 133173294679872 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 287.012, Tokens/s/device: 43261.512, total_weights: 16384, loss: 8.820 I0417 01:15:35.243016 133045419693824 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_09_pdb_lt_1/checkpoints/0/items I0417 01:15:35.244184 133044884641536 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776388534851963647, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776388534851963647, "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_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 01:15:36.381272 2566 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 01:15:38.975472 133044925097728 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_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_0 I0417 01:15:50.080700 133173294679872 metric_logger.py:185] completed step: 4, seconds: 0.054, TFLOP/s/device: 63.265, Tokens/s/device: 9536.049, total_weights: 16384, loss: 7.947 I0417 01:15:50.090714 133173294679872 metric_logger.py:185] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.413, Tokens/s/device: 8804.664, total_weights: 16384, loss: 7.209 I0417 01:15:50.146255 133173294679872 metric_logger.py:185] completed step: 6, seconds: 16.292, TFLOP/s/device: 0.208, Tokens/s/device: 31.426, total_weights: 16384, loss: 6.651 I0417 01:15:50.204981 133173294679872 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 438.238, Tokens/s/device: 66055.993, total_weights: 16384, loss: 6.267 I0417 01:15:50.263725 133173294679872 metric_logger.py:185] completed step: 8, seconds: 0.057, TFLOP/s/device: 59.829, Tokens/s/device: 9018.054, total_weights: 16384, loss: 6.030 I0417 01:15:50.323242 133173294679872 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 01:15:50.323951 133173294679872 checkpointing.py:798] Waited 0.0007178783416748047 seconds for step 9 to finish before starting checkpointing. I0417 01:15:50.325809 133173294679872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 01:16:12.535875 133044916705024 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.162764s (commit=34.932081s, array_metadata_write=1.230683s) I0417 01:16:12.537095 133045962213120 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 39.570 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 39.918704986572266 s) (per-host) I0417 01:16:12.537226 133045962213120 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.645637s. I0417 01:16:13.306639 133045962213120 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776388534851963647, '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_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 01:16:14.055816 133045962213120 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000369s. I0417 01:16:14.058316 133044884641536 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776388534851963647, '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_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0417 01:16:14.680570 133045962213120 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 01:16:15.600392 133045962213120 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.167304s. 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_09_pdb_lt_1/checkpoints/0/items I0417 01:16:15.603084 133045962213120 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_09_pdb_lt_1/checkpoints/0/items I0417 01:16:16.119943 133045962213120 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_09_pdb_lt_1/checkpoints/0 I0417 01:16:17.942190 133045962213120 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_09_pdb_lt_1/checkpoints/0`. I0417 01:16:17.943066 133045962213120 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.05 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_09_pdb_lt_1/checkpoints/0 I0417 01:16:17.944561 133045962213120 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 45.052971s. I0417 01:16:17.944786 133044908312320 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 01:16:17.944934 133044908312320 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 01:16:17.945016 133044908312320 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 01:16:17.946905 133044908312320 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 01:16:17.947049 133173294679872 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 01:16:17.947130 133173294679872 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.621322 seconds. If this number is high, consider checkpointing less frequently. I0417 01:16:17.948932 133173294679872 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 01:16:17.950926 133173294679872 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_09_pdb_lt_1/checkpoints/9. I0417 01:16:18.664527 133173294679872 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 01:16:18.664622 133173294679872 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 01:16:18.691885 133173294679872 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028783s I0417 01:16:18.692637 133173294679872 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.123 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037511587142944336 s) (per-host) I0417 01:16:18.692716 133173294679872 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037609s (batch_requests_ready=0.002835s, total_serialization_initiated=0.034098s, others=0.000677s) I0417 01:16:18.692803 133173294679872 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041522s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.041506) I0417 01:16:18.693523 133173294679872 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_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_09_pdb_lt_1/checkpoints/9. I0417 01:16:18.693866 133044916705024 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:36:18.693812 I0417 01:16:18.696034 133173294679872 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 01:16:18.696307 133044908312320 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 01:16:18.696438 133173294679872 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_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776388550.325777, 'wait_for_prev_duration_secs': 27.6213219165802, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776388577.9489703, 'checkpointer_blocking_duration_secs': 0.7450368404388428, 'get_old_steps_start_time': 1776388578.6940322, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776388550.3241758, 'checkpoint_manager_blocking_duration_secs': 28.372227668762207} I0417 01:16:18.696549 133173294679872 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 01:16:18.696594 133173294679872 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 01:16:18.904537 133045962213120 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_09_pdb_lt_1/checkpoints/9 I0417 01:16:21.045089 133044379834112 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_09_pdb_lt_1/checkpoints/9/items I0417 01:16:25.058291 133045953820416 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_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_0 I0417 01:17:03.561481 133044925097728 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.670543s (commit=39.377371s, array_metadata_write=1.293172s) I0417 01:17:03.562582 133044916705024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.175 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.90744662284851 s) (per-host) I0417 01:17:03.562660 133044916705024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.868680s. I0417 01:17:05.519304 133044916705024 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000531s. I0417 01:17:05.621148 133044916705024 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 01:17:06.563366 133044916705024 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.180067s. 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_09_pdb_lt_1/checkpoints/9/items I0417 01:17:06.565144 133044916705024 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_09_pdb_lt_1/checkpoints/9/items I0417 01:17:07.101613 133044916705024 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_09_pdb_lt_1/checkpoints/9 I0417 01:17:08.947016 133044916705024 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_09_pdb_lt_1/checkpoints/9`. I0417 01:17:08.947814 133044916705024 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.00 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_09_pdb_lt_1/checkpoints/9 I0417 01:17:08.949429 133044916705024 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.255445s. I0417 01:17:08.949605 133044908312320 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 01:17:08.949721 133044908312320 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 01:17:08.949786 133044908312320 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 01:17:08.951552 133044908312320 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 01:17:08.953362 133173294679872 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 01:17:08.953480 133173294679872 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 01:17:08.954230 133173294679872 metric_logger.py:185] completed step: 9, seconds: 0.058, TFLOP/s/device: 58.441, Tokens/s/device: 8808.905, 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: Fri Apr 17 01:17:18 UTC 2026 EXIT_CODE=0