2026-04-20 16:11:49.429854: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0420 16:11:49.551655 132000386153600 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0420 16:12:20.150269 132000386153600 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 16:12:20.150392 132000386153600 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 16:12:20.150426 132000386153600 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0420 16:12:20.150453 132000386153600 train_utils.py:365] WARNING: 'dataset_path' might be pointing your local file system I0420 16:12:20.150544 132000386153600 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0420 16:12:20.668912 132000386153600 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 16:12:21.125508 132000386153600 checkpointing.py:688] Setting up checkpoint logger... I0420 16:12:21.125618 132000386153600 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 16:12:21.125655 132000386153600 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0420 16:12:21.125837 132000386153600 base_pytree_checkpoint_handler.py:411] 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 0x780d25301a00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 16:12:23.518428 132000386153600 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 16:12:23.518621 132000386153600 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7807795c82c0>}, handler_registry=None I0420 16:12:23.518836 132000386153600 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7807795c82c0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0420 16:12:23.518878 132000386153600 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78082070a780>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0420 16:12:23.518907 132000386153600 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 0x7807795c82c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7807795c82c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78082070a780>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78082070a780>}). I0420 16:12:23.519220 132000386153600 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0420 16:12:23.519271 132000386153600 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7806a95b2de0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 16:12:23.653093 132000386153600 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints I0420 16:12:23.653346 132000386153600 checkpoint_manager.py:921] [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_hns=False, 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), root_directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x780820712300> I0420 16:12:23.653425 132000386153600 checkpointing.py:302] Checkpoint manager created! I0420 16:12:23.730624 132000386153600 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0420 16:12:23.733552 132000386153600 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0420 16:12:23.785250 132000386153600 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0420 16:12:23.809184 132000386153600 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 16:12:23.809242 132000386153600 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 16:12:24.456783 132000386153600 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 16:12:24.456900 132000386153600 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0420 16:12:26.648628 132000386153600 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 16:12:26.648719 132000386153600 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 16:12:26.654296 132000386153600 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 16:12:26.654346 132000386153600 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 16:12:26.669398 132000386153600 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 16:12:26.669451 132000386153600 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 16:12:26.683933 132000386153600 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 16:12:26.683987 132000386153600 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 16:12:26.706155 132000386153600 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:12:26.706215 132000386153600 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 16:12:26.720715 132000386153600 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:12:26.720776 132000386153600 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 16:12:26.735278 132000386153600 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:12:26.735333 132000386153600 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 16:12:26.763701 132000386153600 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 16:12:26.763765 132000386153600 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 16:12:26.782998 132000386153600 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 16:12:26.783071 132000386153600 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0420 16:12:34.463615 132000386153600 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0420 16:12:34.464618 132000386153600 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0420 16:12:34.466473 132000386153600 metric_logger.py:301] number parameters: 1.104 billion I0420 16:12:45.165490 132000386153600 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 16:12:45.296162 132000386153600 checkpointing.py:798] Waited 0.13065433502197266 seconds for step 0 to finish before starting checkpointing. I0420 16:12:45.296663 132000386153600 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 16:12:45.296830 132000386153600 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0420 16:12:45.297259 132000386153600 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0. I0420 16:12:45.389893 132000386153600 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0420 16:12:45.504428 132000386153600 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0420 16:12:45.504534 132000386153600 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0420 16:12:45.512099 131885462193728 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 16:12:46.234402 131885451707968 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 16:12:46.380410 131885411862080 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701566145659372, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776701566145659372, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 16:12:46.771459 1163731 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0420 16:12:46.771520 1163731 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0420 16:12:47.181246 132000386153600 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.677924s I0420 16:12:47.186138 132000386153600 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.877 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0420 16:12:47.186203 132000386153600 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.794555s (batch_requests_ready=0.103519s, total_serialization_initiated=1.686265s, others=0.004771s) I0420 16:12:47.186276 132000386153600 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.795804s (all_items=0.000030s, per_item={'items': '0.00003004'}, temp_paths=1.795774) I0420 16:12:47.189530 131885323781696 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 16:12:47.189988 132000386153600 async_checkpointer.py:561] Finished blocking save. Time taken: 1.893116s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0. I0420 16:12:47.190138 132000386153600 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 16:12:47.190303 131885472679488 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 16:12:47.190372 132000386153600 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776701565.2966466, 'wait_for_prev_duration_secs': 5.0067901611328125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776701565.2968538, 'checkpointer_blocking_duration_secs': 1.8932101726531982, 'get_old_steps_start_time': 1776701567.1900833, 'get_old_steps_duration_secs': 2.1219253540039062e-05, 'checkpoint_manager_blocking_start_time': 1776701565.2965548, 'checkpoint_manager_blocking_duration_secs': 1.893791913986206} I0420 16:12:47.190505 132000386153600 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 16:12:47.190555 132000386153600 max_utils.py:750] Memstats: After params initialized: I0420 16:12:47.190602 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0420 16:12:47.190627 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0420 16:12:47.190647 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0420 16:12:47.190668 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0420 16:12:47.190687 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0420 16:12:47.190704 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0420 16:12:47.190726 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0420 16:12:47.190744 132000386153600 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0420 16:12:47.679577 132000386153600 metric_logger.py:196] completed step: 0, seconds: 10.697, TFLOP/s/device: 1.270, Tokens/s/device: 191.447, total_weights: 13328, loss: 10.880, lm_loss: 10.880, perplexity: 53108.402 I0420 16:12:47.680451 132000386153600 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/tensorboard/' I0420 16:12:47.791942 132000386153600 metric_logger.py:196] completed step: 1, seconds: 2.509, TFLOP/s/device: 5.415, Tokens/s/device: 816.147, total_weights: 12332, loss: 10.862, lm_loss: 10.862, perplexity: 52152.828 I0420 16:12:47.908676 132000386153600 metric_logger.py:196] completed step: 2, seconds: 0.022, TFLOP/s/device: 605.352, Tokens/s/device: 91245.266, total_weights: 15161, loss: 9.926, lm_loss: 9.926, perplexity: 20464.070 I0420 16:12:47.996340 131885430736448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0420 16:12:48.026113 132000386153600 metric_logger.py:196] completed step: 3, seconds: 0.107, TFLOP/s/device: 126.798, Tokens/s/device: 19112.322, total_weights: 13327, loss: 9.396, lm_loss: 9.396, perplexity: 12037.986 I0420 16:13:07.857993 132000386153600 metric_logger.py:196] completed step: 4, seconds: 0.115, TFLOP/s/device: 118.457, Tokens/s/device: 17855.119, total_weights: 11939, loss: 9.005, lm_loss: 9.005, perplexity: 8141.998 I0420 16:13:07.871643 132000386153600 metric_logger.py:196] completed step: 5, seconds: 0.117, TFLOP/s/device: 115.648, Tokens/s/device: 17431.716, total_weights: 15502, loss: 8.861, lm_loss: 8.861, perplexity: 7054.059 I0420 16:13:07.985866 132000386153600 metric_logger.py:196] completed step: 6, seconds: 19.833, TFLOP/s/device: 0.685, Tokens/s/device: 103.263, total_weights: 13864, loss: 8.738, lm_loss: 8.738, perplexity: 6233.198 I0420 16:13:08.103179 132000386153600 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1563.176, Tokens/s/device: 235618.960, total_weights: 12988, loss: 8.651, lm_loss: 8.651, perplexity: 5718.487 I0420 16:13:08.220584 132000386153600 metric_logger.py:196] completed step: 8, seconds: 0.116, TFLOP/s/device: 116.983, Tokens/s/device: 17632.979, total_weights: 13820, loss: 8.671, lm_loss: 8.671, perplexity: 5832.209 I0420 16:13:08.338002 132000386153600 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 16:13:08.339240 132000386153600 checkpointing.py:798] Waited 0.0012521743774414062 seconds for step 9 to finish before starting checkpointing. I0420 16:13:08.340245 132000386153600 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 16:13:12.201858 131885441222208 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 25.012989s (commit=24.551548s, array_metadata_write=0.461441s) I0420 16:13:12.203150 131885323781696 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 471.320 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 26 seconds) (per-host) I0420 16:13:12.203206 131885323781696 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.013258s. I0420 16:13:12.444057 131885323781696 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701566145659372, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 16:13:12.621271 131885323781696 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0420 16:13:12.822375 131885411862080 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701566145659372, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 16:13:13.057663 131885323781696 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 16:13:13.057902 131885323781696 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.572746s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 16:13:13.058610 131885323781696 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 16:13:13.301823 131885323781696 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 16:13:14.000009 131885323781696 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0`. I0420 16:13:14.000760 131885323781696 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 28.703895s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 16:13:14.000841 131885323781696 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 26.810894s. I0420 16:13:14.000982 131885472679488 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 16:13:14.001035 131885472679488 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 16:13:14.001100 131885472679488 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 16:13:14.001142 131885472679488 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 16:13:14.001291 132000386153600 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 16:13:14.001451 132000386153600 checkpoint_manager.py:1441] Waiting for previous save to complete took 5.661212 seconds. If this number is high, consider checkpointing less frequently. I0420 16:13:14.002244 132000386153600 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0420 16:13:14.002590 132000386153600 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9. I0420 16:13:14.173147 131885472679488 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 16:13:14.189620 132000386153600 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0420 16:13:14.189730 132000386153600 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0420 16:13:14.902405 131885344753216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 16:13:19.530097 132000386153600 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.341538s I0420 16:13:19.537627 132000386153600 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.267 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0420 16:13:19.537711 132000386153600 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.442641s (batch_requests_ready=0.085661s, total_serialization_initiated=5.349590s, others=0.007390s) I0420 16:13:19.537814 132000386153600 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.443289s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=5.443274) I0420 16:13:19.539683 131885430736448 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 16:13:19.539807 132000386153600 async_checkpointer.py:561] Finished blocking save. Time taken: 5.537507s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9. I0420 16:13:19.540120 132000386153600 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 16:13:19.540338 131885323781696 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 16:13:19.540478 132000386153600 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776701588.3401978, 'wait_for_prev_duration_secs': 5.661211729049683, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776701594.0022714, 'checkpointer_blocking_duration_secs': 5.5376856327056885, 'get_old_steps_start_time': 1776701599.5399828, 'get_old_steps_duration_secs': 8.106231689453125e-05, 'checkpoint_manager_blocking_start_time': 1776701588.3401566, 'checkpoint_manager_blocking_duration_secs': 11.200285911560059} I0420 16:13:19.540635 132000386153600 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 16:13:19.540676 132000386153600 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 16:13:20.301728 131885451707968 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0420 16:13:58.232220 131885344753216 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.693522s (commit=38.240392s, array_metadata_write=0.453130s) I0420 16:13:58.234579 131885430736448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 286.293 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0420 16:13:58.234668 131885430736448 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.694800s. I0420 16:13:58.752581 131885430736448 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0420 16:13:59.146878 131885430736448 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 16:13:59.147144 131885430736448 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.550887s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 16:13:59.147964 131885430736448 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 16:13:59.399970 131885430736448 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 16:14:00.084711 131885430736448 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9`. I0420 16:14:00.085495 131885430736448 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.083207s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 16:14:00.085626 131885430736448 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.545765s. I0420 16:14:00.085909 131885323781696 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 16:14:00.086051 131885323781696 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 16:14:00.086194 131885323781696 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 16:14:00.086266 131885323781696 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 16:14:00.086553 132000386153600 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 16:14:00.086805 132000386153600 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 16:14:00.088670 132000386153600 metric_logger.py:196] completed step: 9, seconds: 0.117, TFLOP/s/device: 116.295, Tokens/s/device: 17529.294, total_weights: 12300, loss: 8.668, lm_loss: 8.668, perplexity: 5810.989 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops