2026-04-20 15:40:12.098929: 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 15:40:12.216209 139444990774400 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0420 15:41:04.331197 139444990774400 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 15:41:04.331320 139444990774400 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 15:41:04.331358 139444990774400 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 15:41:04.331389 139444990774400 train_utils.py:365] WARNING: 'dataset_path' might be pointing your local file system I0420 15:41:04.331492 139444990774400 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0420 15:41:04.835932 139444990774400 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 15:41:05.304842 139444990774400 checkpointing.py:688] Setting up checkpoint logger... I0420 15:41:05.304939 139444990774400 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 15:41:05.304978 139444990774400 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0420 15:41:05.305178 139444990774400 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 0x7ed27a300f50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 15:41:08.044957 139444990774400 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 15:41:08.045175 139444990774400 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 0x7ecd7bf45be0>}, handler_registry=None I0420 15:41:08.045404 139444990774400 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ecd7bf45be0>` 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 15:41:08.045445 139444990774400 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ecd75771ee0>` 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 15:41:08.045474 139444990774400 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 0x7ecd7bf45be0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ecd7bf45be0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ecd75771ee0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ecd75771ee0>}). I0420 15:41:08.045790 139444990774400 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0420 15:41:08.045840 139444990774400 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7ecbff3b2de0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 15:41:08.164618 139444990774400 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_01_base/checkpoints I0420 15:41:08.164839 139444990774400 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_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ecd75770500> I0420 15:41:08.164918 139444990774400 checkpointing.py:302] Checkpoint manager created! I0420 15:41:08.242685 139444990774400 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0420 15:41:08.245499 139444990774400 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 15:41:08.303712 139444990774400 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 15:41:08.327378 139444990774400 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 15:41:08.327442 139444990774400 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 15:41:08.971614 139444990774400 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 15:41:08.971722 139444990774400 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 15:41:11.191036 139444990774400 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 15:41:11.191129 139444990774400 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 15:41:11.196480 139444990774400 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 15:41:11.196528 139444990774400 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 15:41:11.211709 139444990774400 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 15:41:11.211758 139444990774400 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 15:41:11.226169 139444990774400 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 15:41:11.226217 139444990774400 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 15:41:11.248224 139444990774400 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 15:41:11.248282 139444990774400 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 15:41:11.262820 139444990774400 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 15:41:11.262868 139444990774400 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 15:41:11.277378 139444990774400 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 15:41:11.277429 139444990774400 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 15:41:11.306669 139444990774400 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 15:41:11.306730 139444990774400 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 15:41:11.324884 139444990774400 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 15:41:11.324938 139444990774400 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0420 15:41:18.904760 139444990774400 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 15:41:18.905508 139444990774400 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0420 15:41:18.907283 139444990774400 metric_logger.py:301] number parameters: 1.104 billion I0420 15:41:29.855849 139444990774400 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 15:41:29.987686 139444990774400 checkpointing.py:798] Waited 0.1318347454071045 seconds for step 0 to finish before starting checkpointing. I0420 15:41:29.988459 139444990774400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 15:41:29.988661 139444990774400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0420 15:41:29.989171 139444990774400 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_01_base/checkpoints/0. I0420 15:41:30.088249 139444990774400 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0420 15:41:30.184145 139330047706688 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_01_base/checkpoints/0 I0420 15:41:30.189458 139444990774400 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0420 15:41:30.189560 139444990774400 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 15:41:30.923793 139330037220928 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_01_base/checkpoints/0/items I0420 15:41:31.068149 139329997375040 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776699690830978228, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776699690830978228, "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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0420 15:41:31.461298 139444990774400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.273033s I0420 15:41:31.464594 1123840 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0420 15:41:31.464657 1123840 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0420 15:41:31.468345 139444990774400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 8.953 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0420 15:41:31.468409 139444990774400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.378520s (batch_requests_ready=0.089135s, total_serialization_initiated=1.288743s, others=0.000643s) I0420 15:41:31.468482 139444990774400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.379715s (all_items=0.000027s, per_item={'items': '0.00002694'}, temp_paths=1.379688) I0420 15:41:31.470738 139329903003200 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 15:41:31.470838 139444990774400 async_checkpointer.py:561] Finished blocking save. Time taken: 1.482133s. 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_01_base/checkpoints/0. I0420 15:41:31.471346 139444990774400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 15:41:31.471538 139330058192448 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 15:41:31.471934 139444990774400 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776699689.9884431, 'wait_for_prev_duration_secs': 5.8650970458984375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776699689.988684, 'checkpointer_blocking_duration_secs': 1.4825727939605713, 'get_old_steps_start_time': 1776699691.4712775, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776699689.9883316, 'checkpoint_manager_blocking_duration_secs': 1.4835741519927979} I0420 15:41:31.472064 139444990774400 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 15:41:31.472126 139444990774400 max_utils.py:750] Memstats: After params initialized: I0420 15:41:31.472177 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0420 15:41:31.472204 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0420 15:41:31.472225 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0420 15:41:31.472246 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0420 15:41:31.472266 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0420 15:41:31.472285 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0420 15:41:31.472303 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0420 15:41:31.472321 139444990774400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0420 15:41:31.905889 139444990774400 metric_logger.py:196] completed step: 0, seconds: 10.947, TFLOP/s/device: 1.241, Tokens/s/device: 187.084, total_weights: 13328, loss: 10.880, lm_loss: 10.880, perplexity: 53108.402 I0420 15:41:31.906895 139444990774400 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_01_base/tensorboard/' I0420 15:41:32.018251 139444990774400 metric_logger.py:196] completed step: 1, seconds: 2.045, TFLOP/s/device: 6.643, Tokens/s/device: 1001.292, total_weights: 12332, loss: 10.862, lm_loss: 10.862, perplexity: 52152.828 I0420 15:41:32.134947 139444990774400 metric_logger.py:196] completed step: 2, seconds: 0.023, TFLOP/s/device: 599.423, Tokens/s/device: 90351.612, total_weights: 15161, loss: 9.926, lm_loss: 9.926, perplexity: 20464.070 I0420 15:41:32.207676 139329944946240 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_01_base/checkpoints/0/items/array_metadatas/process_0 I0420 15:41:32.252765 139444990774400 metric_logger.py:196] completed step: 3, seconds: 0.105, TFLOP/s/device: 129.506, Tokens/s/device: 19520.564, total_weights: 13327, loss: 9.396, lm_loss: 9.396, perplexity: 12037.986 I0420 15:41:52.027536 139444990774400 metric_logger.py:196] completed step: 4, seconds: 0.117, TFLOP/s/device: 116.481, Tokens/s/device: 17557.245, total_weights: 11939, loss: 9.005, lm_loss: 9.005, perplexity: 8141.998 I0420 15:41:52.041360 139444990774400 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.202, Tokens/s/device: 17364.467, total_weights: 15502, loss: 8.861, lm_loss: 8.861, perplexity: 7054.059 I0420 15:41:52.155962 139444990774400 metric_logger.py:196] completed step: 6, seconds: 19.776, TFLOP/s/device: 0.687, Tokens/s/device: 103.561, total_weights: 13864, loss: 8.738, lm_loss: 8.738, perplexity: 6233.198 I0420 15:41:52.273750 139444990774400 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1529.910, Tokens/s/device: 230604.662, total_weights: 12988, loss: 8.651, lm_loss: 8.651, perplexity: 5718.487 I0420 15:41:52.391092 139444990774400 metric_logger.py:196] completed step: 8, seconds: 0.116, TFLOP/s/device: 117.535, Tokens/s/device: 17716.110, total_weights: 13820, loss: 8.671, lm_loss: 8.671, perplexity: 5832.209 I0420 15:41:52.509332 139444990774400 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 15:41:52.510568 139444990774400 checkpointing.py:798] Waited 0.0012481212615966797 seconds for step 9 to finish before starting checkpointing. I0420 15:41:52.510850 139444990774400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 15:42:06.583203 139329923974720 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.114381s (commit=34.633075s, array_metadata_write=0.481306s) I0420 15:42:06.584430 139329903003200 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 346.266 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0420 15:42:06.584535 139329903003200 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.113563s. I0420 15:42:06.819240 139329903003200 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776699690830978228, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0420 15:42:07.012911 139329903003200 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 15:42:07.213896 139329997375040 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776699690830978228, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA I0420 15:42:07.420672 139329903003200 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 15:42:07.420865 139329903003200 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.562498s. 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_01_base/checkpoints/0/items I0420 15:42:07.421598 139329903003200 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_01_base/checkpoints/0/items I0420 15:42:07.671204 139329903003200 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_01_base/checkpoints/0 I0420 15:42:08.336864 139329903003200 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_01_base/checkpoints/0`. I0420 15:42:08.337522 139329903003200 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.348824s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_01_base/checkpoints/0 I0420 15:42:08.337620 139329903003200 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.866650s. I0420 15:42:08.337744 139330058192448 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 15:42:08.337796 139330058192448 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 15:42:08.337846 139330058192448 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 15:42:08.337889 139330058192448 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 15:42:08.338011 139444990774400 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 15:42:08.338154 139444990774400 checkpoint_manager.py:1441] Waiting for previous save to complete took 15.827307 seconds. If this number is high, consider checkpointing less frequently. I0420 15:42:08.338976 139444990774400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0420 15:42:08.339293 139444990774400 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_01_base/checkpoints/9. I0420 15:42:08.518527 139330058192448 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_01_base/checkpoints/9 I0420 15:42:08.533828 139444990774400 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0420 15:42:08.533924 139444990774400 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 15:42:09.202212 139329892517440 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_01_base/checkpoints/9/items I0420 15:42:13.111951 139444990774400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.579127s I0420 15:42:13.112324 139444990774400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.637 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0420 15:42:13.112432 139444990774400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.679384s (batch_requests_ready=0.092350s, total_serialization_initiated=4.586740s, others=0.000293s) I0420 15:42:13.112498 139444990774400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.680171s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=4.680154) I0420 15:42:13.118890 139329923974720 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 15:42:13.118993 139444990774400 async_checkpointer.py:561] Finished blocking save. Time taken: 4.779961s. 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_01_base/checkpoints/9. I0420 15:42:13.119199 139444990774400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 15:42:13.119344 139329903003200 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 15:42:13.119425 139444990774400 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776699712.5108092, 'wait_for_prev_duration_secs': 15.827306747436523, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776699728.3390114, 'checkpointer_blocking_duration_secs': 4.780107021331787, 'get_old_steps_start_time': 1776699733.1191344, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776699712.5107622, 'checkpoint_manager_blocking_duration_secs': 20.60864019393921} I0420 15:42:13.119506 139444990774400 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 15:42:13.119539 139444990774400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 15:42:13.876647 139329976403520 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_01_base/checkpoints/9/items/array_metadatas/process_0 I0420 15:42:48.148625 139330058192448 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.030452s (commit=34.573485s, array_metadata_write=0.456967s) I0420 15:42:48.149886 139329923974720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 318.173 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0420 15:42:48.150001 139329923974720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.030956s. I0420 15:42:48.585003 139329923974720 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 15:42:49.006256 139329923974720 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 15:42:49.006913 139329923974720 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.560345s. 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_01_base/checkpoints/9/items I0420 15:42:49.007619 139329923974720 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_01_base/checkpoints/9/items I0420 15:42:49.259094 139329923974720 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_01_base/checkpoints/9 I0420 15:42:49.931798 139329923974720 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_01_base/checkpoints/9`. I0420 15:42:49.932468 139329923974720 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.593437s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/nnx_feat_nnx_trainstate_and_training_loop_20260420_142345_01_base/checkpoints/9 I0420 15:42:49.932553 139329923974720 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.813513s. I0420 15:42:49.932708 139329903003200 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 15:42:49.932761 139329903003200 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 15:42:49.932838 139329903003200 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 15:42:49.932886 139329903003200 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 15:42:49.933031 139444990774400 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 15:42:49.933207 139444990774400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 15:42:49.933997 139444990774400 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.206, Tokens/s/device: 17365.056, 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