XPK Start: Sun Apr 19 02:20:09 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 02:20:33.620750: 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) I0419 02:20:33.799084 140577630697280 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 02:20:42.840769 140577630697280 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 02:20:42.843077 140577630697280 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-5202b-slice-job-0-0.mt-15-ocdbt-true-sv-5202b:8482 I0419 02:20:44.932504 140577630697280 max_utils.py:284] Jax distributed system initialized! I0419 02:20:51.218696 140577630697280 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 02:20:51.218796 140577630697280 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 02:20:51.218858 140577630697280 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 I0419 02:20:51.218895 140577630697280 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 02:20:52.254636 140577630697280 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 02:20:52.368113 140577630697280 checkpointing.py:688] Setting up checkpoint logger... I0419 02:20:52.368228 140577630697280 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 02:20:52.368272 140577630697280 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 02:20:52.368473 140577630697280 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 0x7fda1013c470>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 02:20:55.362304 140577630697280 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 02:20:55.362547 140577630697280 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 0x7fd903a6ac30>}, handler_registry=None I0419 02:20:55.362786 140577630697280 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fd903a6ac30>` 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`. I0419 02:20:55.362846 140577630697280 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fc5dc3d1070>` 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`. I0419 02:20:55.362882 140577630697280 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 0x7fd903a6ac30>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fd903a6ac30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fc5dc3d1070>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fc5dc3d1070>}). I0419 02:20:55.363197 140577630697280 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 02:20:55.363266 140577630697280 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7fc5105204a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 02:20:57.328850 140577630697280 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints I0419 02:20:57.331048 140577630697280 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fc5c43c8a40> I0419 02:20:57.331158 140577630697280 checkpointing.py:302] Checkpoint manager created! I0419 02:20:58.276929 140577630697280 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 02:20:58.277044 140577630697280 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 02:20:59.645402 140577630697280 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 02:20:59.645535 140577630697280 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:20:59.645609 140577630697280 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:20:59.645674 140577630697280 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 02:20:59.645735 140577630697280 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:20:59.645789 140577630697280 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:20:59.645860 140577630697280 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:20:59.645916 140577630697280 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 02:20:59.645970 140577630697280 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:20:59.646022 140577630697280 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:20:59.646074 140577630697280 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 02:20:59.646124 140577630697280 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 02:21:00.044320 140577630697280 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:21:00.044411 140577630697280 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:21:00.049735 140577630697280 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:21:00.049791 140577630697280 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:21:00.065903 140577630697280 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:21:00.065963 140577630697280 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:21:00.081552 140577630697280 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:21:00.081608 140577630697280 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:21:00.105268 140577630697280 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:21:00.105339 140577630697280 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:21:00.120911 140577630697280 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:21:00.120969 140577630697280 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:21:00.136470 140577630697280 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:21:00.136528 140577630697280 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:21:00.166551 140577630697280 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 02:21:00.166622 140577630697280 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:21:00.186179 140577630697280 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 02:21:00.186241 140577630697280 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 02:21:11.642575 140577630697280 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. I0419 02:21:11.652890 140577630697280 metric_logger.py:301] number parameters: 1.104 billion I0419 02:21:23.425700 140577630697280 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 02:21:23.578137 140577630697280 checkpointing.py:798] Waited 0.1524207592010498 seconds for step 0 to finish before starting checkpointing. I0419 02:21:23.580630 140577630697280 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:21:23.582375 140577630697280 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0419 02:21:23.583702 140577630697280 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0. I0419 02:21:24.296695 140577630697280 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 02:21:24.316374 140577630697280 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:21:24.316448 140577630697280 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 I0419 02:21:24.583301 140448401626880 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0 I0419 02:21:24.586837 140577630697280 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.271974s I0419 02:21:24.587680 140577630697280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.449 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2831113338470459 s) (per-host) I0419 02:21:24.587754 140577630697280 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283206s (batch_requests_ready=0.003346s, total_serialization_initiated=0.279099s, others=0.000761s) I0419 02:21:24.587852 140577630697280 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287246s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.287229) I0419 02:21:24.588650 140577630697280 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0. I0419 02:21:24.588935 140448410019584 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:41:24.588902 I0419 02:21:24.599487 140577630697280 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:21:24.599790 140445708064512 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:21:24.599934 140577630697280 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565283.5806115, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776565283.5824125, 'checkpointer_blocking_duration_secs': 1.0066659450531006, 'get_old_steps_start_time': 1776565284.5891035, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776565283.5785236, 'checkpoint_manager_blocking_duration_secs': 1.021371841430664} I0419 02:21:24.600096 140577630697280 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 02:21:24.600165 140577630697280 max_utils.py:750] Memstats: After params initialized: I0419 02:21:24.600215 140577630697280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0419 02:21:24.600248 140577630697280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0419 02:21:24.600280 140577630697280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0419 02:21:24.600303 140577630697280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0419 02:21:24.939414 140577630697280 metric_logger.py:196] completed step: 0, seconds: 11.441, TFLOP/s/device: 1.188, Tokens/s/device: 179.007, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:21:24.942417 140577630697280 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/tensorboard/' I0419 02:21:25.337583 140577630697280 metric_logger.py:196] completed step: 1, seconds: 1.510, TFLOP/s/device: 8.998, Tokens/s/device: 1356.285, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:21:25.479453 140577630697280 metric_logger.py:196] completed step: 2, seconds: 0.399, TFLOP/s/device: 34.046, Tokens/s/device: 5131.790, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0419 02:21:25.627298 140577630697280 metric_logger.py:196] completed step: 3, seconds: 0.013, TFLOP/s/device: 1084.714, Tokens/s/device: 163499.920, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0419 02:21:25.639505 140577630697280 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0419 02:21:25.922875 140577630697280 checkpointing.py:798] Waited 0.28331446647644043 seconds for step 5 to finish before starting checkpointing. I0419 02:21:25.925506 140577630697280 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:21:31.902768 140447913604864 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/items I0419 02:21:32.029348 140443052402432 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565291627892306, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776565291627892306, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0419 02:21:33.045068 2851 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 02:21:36.385869 140446790702848 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_0 I0419 02:22:09.670237 140446782310144 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.642652s (commit=35.446757s, array_metadata_write=1.195895s) I0419 02:22:09.671753 140448410019584 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.818 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.36712670326233 s) (per-host) I0419 02:22:09.671981 140448410019584 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.082930s. I0419 02:22:10.396429 140448410019584 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565291627892306, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0419 02:22:11.988111 140443052402432 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565291627892306, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0419 02:22:12.225961 140448410019584 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0419 02:22:12.412566 140448410019584 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000746s. I0419 02:22:13.137586 140448410019584 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.187635s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/items I0419 02:22:13.139616 140448410019584 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0/items I0419 02:22:13.673881 140448410019584 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0 I0419 02:22:15.493278 140448410019584 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0`. I0419 02:22:15.494470 140448410019584 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.91 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/0 I0419 02:22:15.495842 140448410019584 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.906790s. I0419 02:22:15.496141 140445708064512 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:22:15.496301 140445708064512 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:22:15.496408 140445708064512 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:22:15.499187 140445708064512 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 02:22:15.499513 140577630697280 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. W0419 02:22:15.499643 140577630697280 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.574146 seconds. If this number is high, consider checkpointing less frequently. I0419 02:22:15.501375 140577630697280 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0419 02:22:15.503621 140577630697280 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5. I0419 02:22:16.670576 140577630697280 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:22:16.670707 140577630697280 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 I0419 02:22:16.707797 140577630697280 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040182s I0419 02:22:16.708726 140577630697280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 25.760 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.059882164001464844 s) (per-host) I0419 02:22:16.708838 140577630697280 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.060017s (batch_requests_ready=0.006791s, total_serialization_initiated=0.052403s, others=0.000823s) I0419 02:22:16.708966 140577630697280 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.064756s (all_items=0.000025s, per_item={'items': '0.00002503'}, temp_paths=0.064731) I0419 02:22:16.709922 140577630697280 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.21 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5. I0419 02:22:16.710360 140445708064512 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:42:16.710318 I0419 02:22:16.712837 140577630697280 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:22:16.713157 140447877658368 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:22:16.713344 140577630697280 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565285.9254594, 'wait_for_prev_duration_secs': 49.57414627075195, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776565335.5014377, 'checkpointer_blocking_duration_secs': 1.2091188430786133, 'get_old_steps_start_time': 1776565336.7105865, 'get_old_steps_duration_secs': 4.124641418457031e-05, 'checkpoint_manager_blocking_start_time': 1776565285.9233437, 'checkpoint_manager_blocking_duration_secs': 50.789947271347046} I0419 02:22:16.713491 140577630697280 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0419 02:22:16.893064 140448410019584 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5 I0419 02:22:19.388913 140447894443776 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5/items I0419 02:22:22.654655 140447905212160 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_0 I0419 02:22:40.285252 140577630697280 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.947, Tokens/s/device: 14311.470, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0419 02:22:40.296695 140577630697280 metric_logger.py:196] completed step: 5, seconds: 0.148, TFLOP/s/device: 92.106, Tokens/s/device: 13883.146, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0419 02:22:40.440719 140577630697280 metric_logger.py:196] completed step: 6, seconds: 74.657, TFLOP/s/device: 0.182, Tokens/s/device: 27.432, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0419 02:22:40.588198 140577630697280 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1486.394, Tokens/s/device: 224045.509, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0419 02:22:40.735793 140577630697280 metric_logger.py:196] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.227, Tokens/s/device: 14052.229, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0419 02:22:40.884128 140577630697280 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 02:22:40.884833 140577630697280 checkpointing.py:798] Waited 0.0007207393646240234 seconds for step 9 to finish before starting checkpointing. I0419 02:22:40.886864 140577630697280 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:23:03.008446 140447886051072 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.069599s (commit=40.878565s, array_metadata_write=1.191035s) I0419 02:23:03.009561 140445708064512 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.072 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.36070203781128 s) (per-host) I0419 02:23:03.009674 140445708064512 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.299159s. I0419 02:23:05.197702 140445708064512 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0419 02:23:05.266590 140445708064512 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000404s. I0419 02:23:06.256474 140445708064512 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.011731s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5/items I0419 02:23:06.258247 140445708064512 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5/items I0419 02:23:06.806670 140445708064512 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5 I0419 02:23:08.207320 140445708064512 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5`. I0419 02:23:08.208158 140445708064512 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.71 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/5 I0419 02:23:08.209648 140445708064512 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.499130s. I0419 02:23:08.209904 140447877658368 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:23:08.210040 140447877658368 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:23:08.210143 140447877658368 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:23:08.211837 140447877658368 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0419 02:23:08.212015 140577630697280 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. W0419 02:23:08.212145 140577630697280 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.325289 seconds. If this number is high, consider checkpointing less frequently. I0419 02:23:08.213662 140577630697280 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0419 02:23:08.215911 140577630697280 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9. I0419 02:23:08.961514 140577630697280 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:23:08.961610 140577630697280 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 I0419 02:23:08.988380 140577630697280 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028293s I0419 02:23:08.989189 140577630697280 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.653 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037034034729003906 s) (per-host) I0419 02:23:08.989269 140577630697280 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037134s (batch_requests_ready=0.002838s, total_serialization_initiated=0.033569s, others=0.000728s) I0419 02:23:08.989360 140577630697280 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041085s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.041067) I0419 02:23:08.990227 140577630697280 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9. I0419 02:23:08.990514 140447905212160 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:43:08.990475 I0419 02:23:09.004048 140577630697280 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:23:09.004371 140447877658368 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:23:09.004539 140577630697280 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565360.8868303, 'wait_for_prev_duration_secs': 27.325289011001587, 'time_between_consecutive_saves_sec': 25.38757634162903, 'checkpointer_blocking_start_time': 1776565388.2137017, 'checkpointer_blocking_duration_secs': 0.7769513130187988, 'get_old_steps_start_time': 1776565388.990671, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776565360.8850532, 'checkpoint_manager_blocking_duration_secs': 28.11945343017578} I0419 02:23:09.004645 140577630697280 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 02:23:09.004686 140577630697280 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:23:09.197351 140444642023168 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9 I0419 02:23:11.354977 140446802949888 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9/items I0419 02:23:15.269303 140448410019584 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_0 I0419 02:23:54.214741 140447913604864 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.548077s (commit=39.895856s, array_metadata_write=1.652221s) I0419 02:23:54.216053 140447905212160 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.898 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.26386475563049 s) (per-host) I0419 02:23:54.216183 140447905212160 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.225554s. I0419 02:23:56.341413 140447905212160 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0419 02:23:56.437968 140447905212160 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000422s. I0419 02:23:57.375372 140447905212160 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.336819s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9/items I0419 02:23:57.380494 140447905212160 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9/items I0419 02:23:58.332230 140447905212160 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9 I0419 02:23:59.683880 140447905212160 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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9`. I0419 02:23:59.684677 140447905212160 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.47 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_true/checkpoints/9 I0419 02:23:59.686163 140447905212160 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.695531s. I0419 02:23:59.686300 140447877658368 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:23:59.686359 140447877658368 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:23:59.686407 140447877658368 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:23:59.688480 140447877658368 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 02:23:59.688629 140577630697280 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. I0419 02:23:59.688795 140577630697280 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:23:59.689777 140577630697280 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.620, Tokens/s/device: 13960.749, 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: Sun Apr 19 02:24:10 UTC 2026 EXIT_CODE=0