XPK Start: Mon Apr 20 18:26:43 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 18:27:07.558535: 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) I0420 18:27:07.736920 133224282658624 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 18:27:16.778338 133224282658624 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 18:27:16.780699 133224282658624 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-pcbz7-slice-job-0-0.mt-15-ocdbt-true-sv-pcbz7:8482 I0420 18:27:17.788323 133224282658624 max_utils.py:284] Jax distributed system initialized! I0420 18:27:23.782437 133224282658624 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 18:27:23.782537 133224282658624 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 18:27:23.782576 133224282658624 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 I0420 18:27:23.782612 133224282658624 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 18:27:24.472016 133224282658624 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 18:27:24.584765 133224282658624 checkpointing.py:688] Setting up checkpoint logger... I0420 18:27:24.584878 133224282658624 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 18:27:24.584922 133224282658624 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 18:27:24.585128 133224282658624 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 0x7929fa450410>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 18:27:27.606894 133224282658624 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 18:27:27.607148 133224282658624 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 0x7914e87a7860>}, handler_registry=None I0420 18:27:27.607385 133224282658624 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7914e87a7860>` 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`. I0420 18:27:27.607434 133224282658624 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x791608051b80>` 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`. I0420 18:27:27.607472 133224282658624 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 0x7914e87a7860>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7914e87a7860>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x791608051b80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x791608051b80>}). I0420 18:27:27.607789 133224282658624 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 18:27:27.607866 133224282658624 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7915d8494fe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 18:27:29.157629 133224282658624 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints I0420 18:27:30.022944 133224282658624 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7914e87964e0> I0420 18:27:30.023133 133224282658624 checkpointing.py:302] Checkpoint manager created! I0420 18:27:30.966127 133224282658624 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 18:27:30.966240 133224282658624 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 18:27:32.330111 133224282658624 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 18:27:32.330227 133224282658624 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:27:32.330279 133224282658624 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 18:27:32.330323 133224282658624 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 18:27:32.330361 133224282658624 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:27:32.330396 133224282658624 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 18:27:32.330433 133224282658624 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:27:32.330468 133224282658624 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 18:27:32.330502 133224282658624 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:27:32.330534 133224282658624 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 18:27:32.330566 133224282658624 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 18:27:32.330597 133224282658624 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 18:27:32.729747 133224282658624 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:27:32.729839 133224282658624 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:27:32.735225 133224282658624 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:27:32.735279 133224282658624 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:27:32.751345 133224282658624 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:27:32.751403 133224282658624 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:27:32.766901 133224282658624 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:27:32.766958 133224282658624 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 18:27:32.791774 133224282658624 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:27:32.791843 133224282658624 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:27:32.807490 133224282658624 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:27:32.807551 133224282658624 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:27:32.823305 133224282658624 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:27:32.823364 133224282658624 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:27:32.853556 133224282658624 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 18:27:32.853636 133224282658624 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 18:27:32.873549 133224282658624 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 18:27:32.873612 133224282658624 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 18:27:44.128868 133224282658624 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. I0420 18:27:44.138110 133224282658624 metric_logger.py:301] number parameters: 1.104 billion I0420 18:27:55.526571 133224282658624 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 18:27:55.677646 133224282658624 checkpointing.py:798] Waited 0.15104913711547852 seconds for step 0 to finish before starting checkpointing. I0420 18:27:55.680088 133224282658624 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:27:55.681701 133224282658624 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 18:27:55.683219 133224282658624 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0. I0420 18:27:56.037565 133224282658624 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 18:27:56.057701 133224282658624 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:27:56.057770 133224282658624 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 I0420 18:27:56.319594 133096531814144 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0 I0420 18:27:56.333661 133224282658624 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277389s I0420 18:27:56.334487 133224282658624 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.342 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2887394428253174 s) (per-host) I0420 18:27:56.334564 133224282658624 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288836s (batch_requests_ready=0.003345s, total_serialization_initiated=0.284740s, others=0.000751s) I0420 18:27:56.334652 133224282658624 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.293204s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.293186) I0420 18:27:56.335473 133224282658624 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.65 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0. I0420 18:27:56.335806 133096540206848 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:47:56.335767 I0420 18:27:56.342825 133224282658624 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:27:56.343091 133094891063040 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:27:56.343223 133224282658624 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776709675.6800516, 'wait_for_prev_duration_secs': 0.00010251998901367188, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776709675.6817586, 'checkpointer_blocking_duration_secs': 0.654219388961792, 'get_old_steps_start_time': 1776709676.3360038, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776709675.6780565, 'checkpoint_manager_blocking_duration_secs': 0.6651303768157959} I0420 18:27:56.343336 133224282658624 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 18:27:56.343403 133224282658624 max_utils.py:750] Memstats: After params initialized: I0420 18:27:56.343452 133224282658624 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 18:27:56.343484 133224282658624 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 18:27:56.343510 133224282658624 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 18:27:56.343534 133224282658624 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 18:27:56.691419 133224282658624 metric_logger.py:196] completed step: 0, seconds: 11.049, TFLOP/s/device: 1.230, Tokens/s/device: 185.358, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0420 18:27:56.694747 133224282658624 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/tensorboard/' I0420 18:27:57.128905 133224282658624 metric_logger.py:196] completed step: 1, seconds: 1.161, TFLOP/s/device: 11.701, Tokens/s/device: 1763.682, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0420 18:27:57.270619 133224282658624 metric_logger.py:196] completed step: 2, seconds: 0.438, TFLOP/s/device: 30.992, Tokens/s/device: 4671.458, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0420 18:27:57.418309 133224282658624 metric_logger.py:196] completed step: 3, seconds: 0.013, TFLOP/s/device: 1063.322, Tokens/s/device: 160275.473, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0420 18:27:57.430721 133224282658624 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0420 18:27:57.713015 133224282658624 checkpointing.py:798] Waited 0.28226542472839355 seconds for step 5 to finish before starting checkpointing. I0420 18:27:57.716102 133224282658624 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:27:58.166195 133096515028736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/items I0420 18:27:58.310343 133092215875328 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776709677913642287, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776709677913642287, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:27:59.326360 2802 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 18:28:01.122166 133095945025280 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_0 I0420 18:28:36.699192 133095437100800 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.377923s (commit=36.114444s, array_metadata_write=1.263480s) I0420 18:28:36.700711 133096540206848 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 38.854 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.65490436553955 s) (per-host) I0420 18:28:36.700926 133096540206848 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.364996s. I0420 18:28:37.887529 133096540206848 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776709677913642287, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:28:38.722820 133092215875328 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776709677913642287, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0420 18:28:39.360478 133096540206848 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:28:39.473510 133096540206848 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000768s. I0420 18:28:40.419385 133096540206848 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.374576s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/items I0420 18:28:40.421569 133096540206848 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0/items I0420 18:28:40.961634 133096540206848 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0 I0420 18:28:42.369606 133096540206848 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0`. I0420 18:28:42.370738 133096540206848 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 46.69 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/0 I0420 18:28:42.372527 133096540206848 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 46.036584s. I0420 18:28:42.372928 133094891063040 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:28:42.373109 133094891063040 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:28:42.373220 133094891063040 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:28:42.375758 133094891063040 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 18:28:42.376120 133224282658624 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. W0420 18:28:42.376308 133224282658624 checkpoint_manager.py:1452] Waiting for previous save to complete took 44.660227 seconds. If this number is high, consider checkpointing less frequently. I0420 18:28:42.378238 133224282658624 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0420 18:28:42.380672 133224282658624 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5. I0420 18:28:43.116945 133224282658624 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:28:43.117095 133224282658624 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 I0420 18:28:43.157056 133224282658624 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043149s I0420 18:28:43.158090 133224282658624 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 24.482 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06300783157348633 s) (per-host) I0420 18:28:43.158189 133224282658624 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.063140s (batch_requests_ready=0.006815s, total_serialization_initiated=0.055420s, others=0.000905s) I0420 18:28:43.158314 133224282658624 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.068193s (all_items=0.000024s, per_item={'items': '0.00002432'}, temp_paths=0.068169) I0420 18:28:43.159348 133224282658624 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5. I0420 18:28:43.159729 133094891063040 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:48:43.159691 I0420 18:28:43.162251 133224282658624 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:28:43.162625 133094367893248 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:28:43.162882 133224282658624 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776709677.7160378, 'wait_for_prev_duration_secs': 44.66022706031799, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776709722.378303, 'checkpointer_blocking_duration_secs': 0.7816066741943359, 'get_old_steps_start_time': 1776709723.159937, 'get_old_steps_duration_secs': 4.124641418457031e-05, 'checkpoint_manager_blocking_start_time': 1776709677.7135973, 'checkpoint_manager_blocking_duration_secs': 45.449233293533325} I0420 18:28:43.163029 133224282658624 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0420 18:28:43.344510 133096540206848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5 I0420 18:28:46.451198 133095945025280 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5/items I0420 18:28:50.131291 133094384678656 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_0 I0420 18:29:07.149825 133224282658624 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.842, Tokens/s/device: 14295.586, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0420 18:29:07.168366 133224282658624 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.481, Tokens/s/device: 13939.749, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0420 18:29:07.306107 133224282658624 metric_logger.py:196] completed step: 6, seconds: 69.737, TFLOP/s/device: 0.195, Tokens/s/device: 29.367, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0420 18:29:07.453501 133224282658624 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1431.882, Tokens/s/device: 215828.854, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0420 18:29:07.600937 133224282658624 metric_logger.py:196] completed step: 8, seconds: 0.140, TFLOP/s/device: 97.383, Tokens/s/device: 14678.689, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0420 18:29:07.749071 133224282658624 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 18:29:07.749791 133224282658624 checkpointing.py:798] Waited 0.0007328987121582031 seconds for step 9 to finish before starting checkpointing. I0420 18:29:07.751816 133224282658624 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:29:29.695741 133094376285952 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.348470s (commit=40.145944s, array_metadata_write=1.202526s) I0420 18:29:29.696921 133094891063040 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.896 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.60182857513428 s) (per-host) I0420 18:29:29.697053 133094891063040 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.537182s. I0420 18:29:31.125842 133094891063040 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000314s. I0420 18:29:31.343949 133094891063040 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:29:32.459611 133094891063040 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.950656s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5/items I0420 18:29:32.461305 133094891063040 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5/items I0420 18:29:33.488925 133094891063040 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5 I0420 18:29:35.325036 133094891063040 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5`. I0420 18:29:35.325882 133094891063040 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.95 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/5 I0420 18:29:35.327335 133094891063040 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 52.167463s. I0420 18:29:35.327535 133094367893248 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:29:35.327653 133094367893248 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:29:35.327732 133094367893248 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:29:35.330590 133094367893248 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0420 18:29:35.330769 133224282658624 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0420 18:29:35.330899 133224282658624 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.579089 seconds. If this number is high, consider checkpointing less frequently. I0420 18:29:35.332355 133224282658624 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0420 18:29:35.334499 133224282658624 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9. I0420 18:29:35.660851 133224282658624 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:29:35.660949 133224282658624 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 I0420 18:29:35.688286 133224282658624 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028873s I0420 18:29:35.689308 133224282658624 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.983 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03763937950134277 s) (per-host) I0420 18:29:35.689369 133224282658624 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037717s (batch_requests_ready=0.002774s, total_serialization_initiated=0.034016s, others=0.000928s) I0420 18:29:35.689446 133224282658624 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041753s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.041737) I0420 18:29:35.690196 133224282658624 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9. I0420 18:29:35.690531 133095945025280 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:49:35.690492 I0420 18:29:35.702832 133224282658624 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:29:35.703200 133094367893248 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:29:35.703361 133224282658624 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776709747.7517838, 'wait_for_prev_duration_secs': 27.579089164733887, 'time_between_consecutive_saves_sec': 25.375940561294556, 'checkpointer_blocking_start_time': 1776709775.3323956, 'checkpointer_blocking_duration_secs': 0.35828661918640137, 'get_old_steps_start_time': 1776709775.6907015, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776709747.7500281, 'checkpoint_manager_blocking_duration_secs': 27.953301906585693} I0420 18:29:35.703467 133224282658624 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 18:29:35.703509 133224282658624 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:29:35.909953 133094891063040 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9 I0420 18:29:38.624004 133096540206848 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9/items I0420 18:29:42.019356 133096515028736 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_0 I0420 18:30:22.017681 133096506636032 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.214594s (commit=41.017379s, array_metadata_write=1.197215s) I0420 18:30:22.018782 133095945025280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.067 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.36708617210388 s) (per-host) I0420 18:30:22.018887 133095945025280 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.328232s. I0420 18:30:23.801047 133095945025280 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 18:30:24.046728 133095945025280 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000403s. I0420 18:30:24.848578 133095945025280 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.906511s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9/items I0420 18:30:24.850443 133095945025280 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9/items I0420 18:30:25.450421 133095945025280 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9 I0420 18:30:26.929457 133095945025280 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9`. I0420 18:30:26.930286 133095945025280 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.60 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_true/checkpoints/9 I0420 18:30:26.931607 133095945025280 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.240948s. I0420 18:30:26.931828 133094367893248 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:30:26.931943 133094367893248 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:30:26.932002 133094367893248 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:30:26.934751 133094367893248 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 18:30:26.934889 133224282658624 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. I0420 18:30:26.935013 133224282658624 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:30:26.935956 133224282658624 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.212, Tokens/s/device: 13899.258, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 18:30:38 UTC 2026 EXIT_CODE=0