XPK Start: Fri Apr 17 01:26:38 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-17 01:27:02.053253: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0417 01:27:02.583536 135580057163584 max_utils.py:273] Attempting to initialize the jax distributed system... I0417 01:27:11.622901 135580057163584 distributed.py:149] Starting JAX distributed service on [::]:8482 I0417 01:27:11.625270 135580057163584 distributed.py:172] Connecting to JAX distributed service on mt-11-optimizer-offload-lptfc-slice-job-0-0.mt-11-optimizer-offload-lptfc:8482 I0417 01:27:12.679665 135580057163584 max_utils.py:284] Jax distributed system initialized! I0417 01:27:18.871358 135580057163584 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 01:27:18.871463 135580057163584 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 01:27:18.871504 135580057163584 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365 I0417 01:27:18.871538 135580057163584 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 01:27:19.537956 135580057163584 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 01:27:19.651367 135580057163584 checkpointing.py:688] Setting up checkpoint logger... I0417 01:27:19.651483 135580057163584 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 01:27:19.651527 135580057163584 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0417 01:27:19.651725 135580057163584 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 0x7b4e987b17c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 01:27:22.833482 135580057163584 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0417 01:27:22.833711 135580057163584 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 0x7b39dc6b8440>}, handler_registry=None I0417 01:27:22.833994 135580057163584 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b39dc6b8440>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0417 01:27:22.834053 135580057163584 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b39fc540410>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0417 01:27:22.834099 135580057163584 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 0x7b39dc6b8440>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b39dc6b8440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b39fc540410>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b39fc540410>}). I0417 01:27:22.834446 135580057163584 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0417 01:27:22.834518 135580057163584 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b39fc389ee0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0417 01:27:24.684132 135580057163584 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_11_optimizer_offload_false/checkpoints I0417 01:27:25.123298 135580057163584 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b39dc6d3da0> I0417 01:27:25.123469 135580057163584 checkpointing.py:302] Checkpoint manager created! I0417 01:27:26.076058 135580057163584 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0417 01:27:26.076177 135580057163584 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0417 01:27:27.464379 135580057163584 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0417 01:27:27.464516 135580057163584 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 01:27:27.464586 135580057163584 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0417 01:27:27.464649 135580057163584 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0417 01:27:27.464712 135580057163584 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 01:27:27.464768 135580057163584 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0417 01:27:27.464827 135580057163584 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 01:27:27.464903 135580057163584 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0417 01:27:27.464968 135580057163584 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 01:27:27.465028 135580057163584 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0417 01:27:27.465085 135580057163584 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0417 01:27:27.465142 135580057163584 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0417 01:27:27.868808 135580057163584 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 01:27:27.868913 135580057163584 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 01:27:27.874272 135580057163584 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 01:27:27.874326 135580057163584 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 01:27:27.890708 135580057163584 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 01:27:27.890777 135580057163584 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 01:27:27.906526 135580057163584 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 01:27:27.906588 135580057163584 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0417 01:27:27.930661 135580057163584 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 01:27:27.930723 135580057163584 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 01:27:27.946712 135580057163584 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 01:27:27.946773 135580057163584 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 01:27:27.967773 135580057163584 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 01:27:27.967882 135580057163584 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 01:27:27.998229 135580057163584 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 01:27:27.998299 135580057163584 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0417 01:27:28.018188 135580057163584 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 01:27:28.018248 135580057163584 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0417 01:27:39.086386 135580057163584 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 01:27:39.095407 135580057163584 metric_logger.py:289] number parameters: 1.104 billion I0417 01:27:50.554297 135580057163584 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0417 01:27:50.706588 135580057163584 checkpointing.py:798] Waited 0.15227580070495605 seconds for step 0 to finish before starting checkpointing. I0417 01:27:50.709130 135580057163584 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 01:27:50.710960 135580057163584 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0417 01:27:50.712399 135580057163584 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0. I0417 01:27:51.447746 135580057163584 signaling_client.py:364] Using JaxDistributedSignalingClient I0417 01:27:51.467143 135580057163584 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 01:27:51.467213 135580057163584 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 01:27:51.736554 135580057163584 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.270911s I0417 01:27:51.737395 135580057163584 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.473 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2818431854248047 s) (per-host) I0417 01:27:51.737465 135580057163584 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.281931s (batch_requests_ready=0.003040s, total_serialization_initiated=0.278132s, others=0.000758s) I0417 01:27:51.737542 135580057163584 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285805s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.285786) I0417 01:27:51.738360 135580057163584 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 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_11_optimizer_offload_false/checkpoints/0. I0417 01:27:51.738684 135450316502784 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:47:51.738647 I0417 01:27:51.753469 135580057163584 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 01:27:51.753760 135447578339072 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 01:27:51.753917 135580057163584 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776389270.7091115, 'wait_for_prev_duration_secs': 6.0558319091796875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776389270.7109993, 'checkpointer_blocking_duration_secs': 1.0278286933898926, 'get_old_steps_start_time': 1776389271.7388692, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776389270.7070093, 'checkpoint_manager_blocking_duration_secs': 1.046870231628418} I0417 01:27:51.754067 135580057163584 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0417 01:27:51.754142 135580057163584 max_utils.py:750] Memstats: After params initialized: I0417 01:27:51.754194 135580057163584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0417 01:27:51.754226 135580057163584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0417 01:27:51.754253 135580057163584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0417 01:27:51.754276 135580057163584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0417 01:27:52.093606 135580057163584 metric_logger.py:185] completed step: 0, seconds: 11.161, TFLOP/s/device: 1.217, Tokens/s/device: 183.501, total_weights: 65536, loss: 10.874 I0417 01:27:52.096725 135580057163584 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/tensorboard/' I0417 01:27:52.130670 135450308110080 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0 I0417 01:27:52.949960 135580057163584 metric_logger.py:185] completed step: 1, seconds: 1.536, TFLOP/s/device: 8.846, Tokens/s/device: 1333.321, total_weights: 65536, loss: 10.874 I0417 01:27:53.091701 135580057163584 metric_logger.py:185] completed step: 2, seconds: 0.857, TFLOP/s/device: 15.850, Tokens/s/device: 2389.116, total_weights: 65536, loss: 10.262 I0417 01:27:53.239591 135580057163584 metric_logger.py:185] completed step: 3, seconds: 0.012, TFLOP/s/device: 1099.016, Tokens/s/device: 165655.585, total_weights: 65536, loss: 9.732 I0417 01:27:55.197437 135450291324672 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/items I0417 01:27:55.369150 135446519875328 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776389274948381369, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776389274948381369, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 01:27:56.736417 2865 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0417 01:27:58.783297 135449730651904 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0417 01:28:15.875440 135580057163584 metric_logger.py:185] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.962, Tokens/s/device: 14313.771, total_weights: 65536, loss: 9.274 I0417 01:28:15.885935 135580057163584 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.278, Tokens/s/device: 13909.075, total_weights: 65536, loss: 8.888 I0417 01:28:16.030255 135580057163584 metric_logger.py:185] completed step: 6, seconds: 22.634, TFLOP/s/device: 0.600, Tokens/s/device: 90.481, total_weights: 65536, loss: 8.589 I0417 01:28:16.177661 135580057163584 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1696.271, Tokens/s/device: 255680.400, total_weights: 65536, loss: 8.380 I0417 01:28:16.325112 135580057163584 metric_logger.py:185] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.123, Tokens/s/device: 14036.531, total_weights: 65536, loss: 8.251 I0417 01:28:16.473511 135580057163584 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0417 01:28:16.474271 135580057163584 checkpointing.py:798] Waited 0.0007715225219726562 seconds for step 9 to finish before starting checkpointing. I0417 01:28:16.476231 135580057163584 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 01:28:34.135922 135449722259200 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.410017s (commit=35.537120s, array_metadata_write=1.872897s) I0417 01:28:34.137039 135450316502784 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 37.009 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.68146085739136 s) (per-host) I0417 01:28:34.137148 135450316502784 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.398352s. I0417 01:28:34.837272 135450316502784 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776389274948381369, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 01:28:35.574108 135450316502784 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000396s. I0417 01:28:35.838002 135450316502784 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 01:28:36.138571 135446519875328 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776389274948381369, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 01:28:36.903357 135450316502784 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.942558s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/items I0417 01:28:36.905114 135450316502784 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0/items I0417 01:28:37.434951 135450316502784 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0 I0417 01:28:39.296770 135450316502784 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0`. I0417 01:28:39.297595 135450316502784 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 48.59 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/0 I0417 01:28:39.298992 135450316502784 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 47.560193s. I0417 01:28:39.299201 135447578339072 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 01:28:39.299316 135447578339072 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 01:28:39.299380 135447578339072 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 01:28:39.301580 135447578339072 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 01:28:39.301714 135580057163584 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0417 01:28:39.301806 135580057163584 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.825580 seconds. If this number is high, consider checkpointing less frequently. I0417 01:28:39.303823 135580057163584 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0417 01:28:39.306026 135580057163584 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9. I0417 01:28:40.019042 135580057163584 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0417 01:28:40.019143 135580057163584 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0417 01:28:40.045708 135580057163584 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028101s I0417 01:28:40.046625 135580057163584 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 41.567 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.037110328674316406 s) (per-host) I0417 01:28:40.046692 135580057163584 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037194s (batch_requests_ready=0.002638s, total_serialization_initiated=0.033689s, others=0.000867s) I0417 01:28:40.046762 135580057163584 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041241s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.041226) I0417 01:28:40.047420 135580057163584 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9. I0417 01:28:40.047754 135449722259200 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 01:48:40.047715 I0417 01:28:40.060128 135580057163584 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 01:28:40.060382 135447578339072 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 01:28:40.060513 135580057163584 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_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776389296.4762, 'wait_for_prev_duration_secs': 22.82557988166809, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776389319.3038712, 'checkpointer_blocking_duration_secs': 0.7440359592437744, 'get_old_steps_start_time': 1776389320.0479305, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776389296.4744942, 'checkpoint_manager_blocking_duration_secs': 23.585984230041504} I0417 01:28:40.060623 135580057163584 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0417 01:28:40.060666 135580057163584 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 01:28:40.678463 135450316502784 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9 I0417 01:28:42.433915 135428929738496 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9/items I0417 01:28:46.061189 135449751037696 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0417 01:29:25.549593 135449730651904 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.658067s (commit=40.142727s, array_metadata_write=1.515340s) I0417 01:29:25.550794 135449722259200 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.685 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.54125094413757 s) (per-host) I0417 01:29:25.550942 135449722259200 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.503063s. I0417 01:29:27.290515 135449722259200 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0417 01:29:27.460505 135449722259200 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000407s. I0417 01:29:28.332638 135449722259200 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.895873s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9/items I0417 01:29:28.334322 135449722259200 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9/items I0417 01:29:28.883465 135449722259200 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9 I0417 01:29:30.284794 135449722259200 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9`. I0417 01:29:30.285619 135449722259200 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.98 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260417_003418/nnx_xpk_feat_nnx_post_train_fixes_20260417_003418_11_optimizer_offload_false/checkpoints/9 I0417 01:29:30.287872 135449722259200 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.239991s. I0417 01:29:30.288082 135447578339072 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 01:29:30.288215 135447578339072 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 01:29:30.288279 135447578339072 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 01:29:30.291162 135447578339072 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 01:29:30.291333 135580057163584 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0417 01:29:30.291507 135580057163584 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 01:29:30.292185 135580057163584 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.404, Tokens/s/device: 13928.183, total_weights: 65536, loss: 8.175 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 17 01:29:38 UTC 2026 EXIT_CODE=0