XPK Start: Fri Apr 17 02:06:33 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 02:06:57.736299: 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 02:06:57.913023 134859687249728 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 02:07:06.952693 134859687249728 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 02:07:06.955106 134859687249728 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-lgc4r-slice-job-0-0.mt-15-ocdbt-false-sv-lgc4r:8482 I0417 02:07:08.555803 134859687249728 max_utils.py:284] Jax distributed system initialized! I0417 02:07:14.583684 134859687249728 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 02:07:14.583789 134859687249728 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 02:07:14.583845 134859687249728 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 02:07:14.583884 134859687249728 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 02:07:15.260343 134859687249728 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 02:07:15.373323 134859687249728 checkpointing.py:688] Setting up checkpoint logger... I0417 02:07:15.373441 134859687249728 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0417 02:07:15.373485 134859687249728 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 02:07:15.373689 134859687249728 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7aa6e0d18ad0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 02:07:19.088182 134859687249728 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 02:07:19.088412 134859687249728 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a92485a7cb0>}, handler_registry=None I0417 02:07:19.088654 134859687249728 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a92485a7cb0>` 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 02:07:19.088702 134859687249728 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a928021eba0>` 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 02:07:19.088737 134859687249728 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 0x7a92485a7cb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a92485a7cb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a928021eba0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a928021eba0>}). I0417 02:07:19.089078 134859687249728 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 02:07:19.089148 134859687249728 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a92687de020> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 02:07:20.644507 134859687249728 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_15_ocdbt_false/checkpoints I0417 02:07:21.510766 134859687249728 checkpoint_manager.py:929] [process=6][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_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a928012b2c0> I0417 02:07:21.510957 134859687249728 checkpointing.py:302] Checkpoint manager created! I0417 02:07:22.472366 134859687249728 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 02:07:22.472483 134859687249728 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 02:07:23.864016 134859687249728 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 02:07:23.864126 134859687249728 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:07:23.864176 134859687249728 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 02:07:23.864218 134859687249728 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 02:07:23.864256 134859687249728 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:07:23.864289 134859687249728 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 02:07:23.864324 134859687249728 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:07:23.864358 134859687249728 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 02:07:23.864391 134859687249728 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:07:23.864422 134859687249728 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 02:07:23.864453 134859687249728 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 02:07:23.864484 134859687249728 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 02:07:24.261958 134859687249728 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:07:24.262047 134859687249728 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:07:24.267329 134859687249728 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 02:07:24.267385 134859687249728 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:07:24.283635 134859687249728 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:07:24.283695 134859687249728 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:07:24.299566 134859687249728 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 02:07:24.299627 134859687249728 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 02:07:24.323573 134859687249728 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:07:24.323636 134859687249728 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:07:24.339627 134859687249728 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:07:24.339684 134859687249728 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:07:24.355505 134859687249728 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 02:07:24.355572 134859687249728 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:07:24.385605 134859687249728 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 02:07:24.385678 134859687249728 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 02:07:24.405381 134859687249728 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 02:07:24.405441 134859687249728 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 02:07:35.474831 134859687249728 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0417 02:07:35.477213 134859687249728 metric_logger.py:289] number parameters: 1.104 billion I0417 02:07:46.600975 134859687249728 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 02:07:46.753727 134859687249728 checkpointing.py:798] Waited 0.15273809432983398 seconds for step 0 to finish before starting checkpointing. I0417 02:07:46.756762 134859687249728 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:07:46.758350 134859687249728 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0417 02:07:46.759749 134859687249728 event_tracking.py:70] [process=6] [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_15_ocdbt_false/checkpoints/0. I0417 02:07:47.523354 134859687249728 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 02:07:47.524364 134859687249728 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:07:47.524420 134859687249728 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 02:07:47.806252 134859687249728 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.283660s I0417 02:07:47.806425 134859687249728 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.267 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29285645484924316 s) (per-host) I0417 02:07:47.806478 134859687249728 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.292921s (batch_requests_ready=0.003167s, total_serialization_initiated=0.289681s, others=0.000072s) I0417 02:07:47.806571 134859687249728 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.297044s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.297027) I0417 02:07:47.807340 134859687249728 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.05 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_15_ocdbt_false/checkpoints/0. I0417 02:07:47.807688 134730622711552 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:27:47.807650 I0417 02:07:47.810382 134859687249728 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:07:47.810631 134728980141824 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:07:47.810755 134859687249728 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391666.7567432, 'wait_for_prev_duration_secs': 6.699562072753906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776391666.758389, 'checkpointer_blocking_duration_secs': 1.0494706630706787, 'get_old_steps_start_time': 1776391667.8078856, 'get_old_steps_duration_secs': 3.4809112548828125e-05, 'checkpoint_manager_blocking_start_time': 1776391666.754143, 'checkpoint_manager_blocking_duration_secs': 1.0565757751464844} I0417 02:07:47.810875 134859687249728 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 02:07:47.810946 134859687249728 max_utils.py:750] Memstats: After params initialized: I0417 02:07:47.810994 134859687249728 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0417 02:07:47.811025 134859687249728 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0417 02:07:47.811050 134859687249728 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0417 02:07:47.811073 134859687249728 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0417 02:07:48.126801 134859687249728 metric_logger.py:185] completed step: 0, seconds: 11.123, TFLOP/s/device: 1.222, Tokens/s/device: 184.129, total_weights: 65536, loss: 10.874 I0417 02:07:48.301868 134859687249728 metric_logger.py:185] completed step: 1, seconds: 1.524, TFLOP/s/device: 8.914, Tokens/s/device: 1343.583, total_weights: 65536, loss: 10.874 I0417 02:07:48.734401 134859687249728 metric_logger.py:185] completed step: 2, seconds: 0.029, TFLOP/s/device: 467.892, Tokens/s/device: 70525.845, total_weights: 65536, loss: 10.560 I0417 02:07:48.881937 134859687249728 metric_logger.py:185] completed step: 3, seconds: 0.422, TFLOP/s/device: 32.171, Tokens/s/device: 4849.151, total_weights: 65536, loss: 9.979 I0417 02:07:48.887807 134859687249728 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0417 02:07:49.177134 134859687249728 checkpointing.py:798] Waited 0.28929758071899414 seconds for step 5 to finish before starting checkpointing. I0417 02:07:49.179882 134859687249728 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:07:52.377157 2820 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 02:07:55.176945 134728988534528 array_metadata_store.py:203] [process=6][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_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_6 I0417 02:07:59.610738 134730622711552 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 130.576 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 12.097131967544556 s) (per-host) I0417 02:07:59.610911 134730622711552 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.803106s. I0417 02:08:07.168062 134730622711552 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 19.360239s. I0417 02:08:07.168313 134728980141824 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:08:07.168427 134728980141824 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:08:07.168478 134728980141824 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:08:07.170120 134728980141824 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 02:08:07.170315 134859687249728 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0417 02:08:07.170449 134859687249728 checkpoint_manager.py:1452] Waiting for previous save to complete took 17.990565 seconds. If this number is high, consider checkpointing less frequently. I0417 02:08:07.172510 134859687249728 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0417 02:08:07.174998 134859687249728 event_tracking.py:70] [process=6] [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_15_ocdbt_false/checkpoints/5. I0417 02:08:07.892013 134859687249728 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:08:07.892104 134859687249728 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 02:08:07.928155 134859687249728 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037711s I0417 02:08:07.928321 134859687249728 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.476 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04474306106567383 s) (per-host) I0417 02:08:07.928370 134859687249728 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044802s (batch_requests_ready=0.002800s, total_serialization_initiated=0.041936s, others=0.000066s) I0417 02:08:07.928464 134859687249728 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049679s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.049664) I0417 02:08:07.929128 134859687249728 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 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_15_ocdbt_false/checkpoints/5. I0417 02:08:07.929471 134728980141824 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:28:07.929436 I0417 02:08:07.946229 134859687249728 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:08:07.946546 134730605926144 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:08:07.946721 134859687249728 standard_logger.py:34] {'step': 5, '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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391669.1798525, 'wait_for_prev_duration_secs': 17.990565299987793, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776391687.172549, 'checkpointer_blocking_duration_secs': 0.7570745944976807, 'get_old_steps_start_time': 1776391687.9296477, 'get_old_steps_duration_secs': 3.0994415283203125e-05, 'checkpoint_manager_blocking_start_time': 1776391669.177467, 'checkpoint_manager_blocking_duration_secs': 18.769216775894165} I0417 02:08:07.946846 134859687249728 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0417 02:08:07.947712 134859687249728 metric_logger.py:185] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.803, Tokens/s/device: 12481.032, total_weights: 65536, loss: 9.457 I0417 02:08:07.955591 134859687249728 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.371, Tokens/s/device: 13923.164, total_weights: 65536, loss: 8.955 I0417 02:08:14.236992 134728988534528 array_metadata_store.py:203] [process=6][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_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_6 I0417 02:08:19.115996 134728980141824 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 140.629 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.232378959655762 s) (per-host) I0417 02:08:19.116121 134728980141824 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.186536s. I0417 02:08:27.366942 134728980141824 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 19.437340s. I0417 02:08:27.367228 134730605926144 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:08:27.367345 134730605926144 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:08:27.367395 134730605926144 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:08:27.368999 134730605926144 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 02:08:31.679460 134859687249728 metric_logger.py:185] completed step: 6, seconds: 19.066, TFLOP/s/device: 0.713, Tokens/s/device: 107.416, total_weights: 65536, loss: 8.465 I0417 02:08:31.827057 134859687249728 metric_logger.py:185] completed step: 7, seconds: 23.578, TFLOP/s/device: 0.576, Tokens/s/device: 86.859, total_weights: 65536, loss: 7.998 I0417 02:08:31.974706 134859687249728 metric_logger.py:185] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.553, Tokens/s/device: 13347.672, total_weights: 65536, loss: 7.567 I0417 02:08:32.123201 134859687249728 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 02:08:32.123901 134859687249728 checkpointing.py:798] Waited 0.0007135868072509766 seconds for step 9 to finish before starting checkpointing. I0417 02:08:32.125941 134859687249728 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:08:32.127907 134859687249728 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0417 02:08:32.129258 134859687249728 event_tracking.py:70] [process=6] [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_15_ocdbt_false/checkpoints/9. I0417 02:08:32.839925 134859687249728 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 02:08:32.840015 134859687249728 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 02:08:32.876677 134859687249728 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.038414s I0417 02:08:32.876811 134859687249728 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 33.899 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04550504684448242 s) (per-host) I0417 02:08:32.876876 134859687249728 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045579s (batch_requests_ready=0.002957s, total_serialization_initiated=0.042544s, others=0.000078s) I0417 02:08:32.876962 134859687249728 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049796s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.049785) I0417 02:08:32.877581 134859687249728 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 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_15_ocdbt_false/checkpoints/9. I0417 02:08:32.877909 134708117612288 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 02:28:32.877871 I0417 02:08:32.879900 134859687249728 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 02:08:32.880186 134708465755904 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0417 02:08:32.880342 134859687249728 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776391712.1259127, 'wait_for_prev_duration_secs': 6.794929504394531e-05, 'time_between_consecutive_saves_sec': 4.756874084472656, 'checkpointer_blocking_start_time': 1776391712.1279435, 'checkpointer_blocking_duration_secs': 0.7501218318939209, 'get_old_steps_start_time': 1776391712.8780828, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776391712.1241252, 'checkpoint_manager_blocking_duration_secs': 0.7561855316162109} I0417 02:08:32.880454 134859687249728 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 02:08:32.880503 134859687249728 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 02:08:38.293017 134706012092160 array_metadata_store.py:203] [process=6][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_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_6 I0417 02:08:43.911744 134708117612288 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 142.558 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.080398321151733 s) (per-host) I0417 02:08:43.911877 134708117612288 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.033846s. I0417 02:08:50.551938 134708117612288 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.673892s. I0417 02:08:50.552215 134708465755904 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 02:08:50.552333 134708465755904 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0417 02:08:50.552374 134708465755904 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 02:08:50.553850 134708465755904 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 02:08:50.554028 134859687249728 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 02:08:50.554205 134859687249728 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 02:08:50.554884 134859687249728 metric_logger.py:185] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.860, Tokens/s/device: 13846.164, total_weights: 65536, loss: 7.176 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 02:08:58 UTC 2026 EXIT_CODE=0