2026-04-16 20:13:01.903273: 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) I0416 20:13:02.018989 132919762885760 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 20:14:00.272687 132919762885760 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 20:14:00.272798 132919762885760 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 20:14:00.272831 132919762885760 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 I0416 20:14:00.272858 132919762885760 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 20:14:00.272954 132919762885760 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0416 20:14:00.777647 132919762885760 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 20:14:00.875680 132919762885760 checkpointing.py:688] Setting up checkpoint logger... I0416 20:14:00.875778 132919762885760 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 20:14:00.875826 132919762885760 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 20:14:00.876009 132919762885760 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 0x78e334ffcf80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 20:14:03.373583 132919762885760 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 20:14:03.373773 132919762885760 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 0x78dcb5398da0>}, handler_registry=None I0416 20:14:03.374073 132919762885760 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78dcb5398da0>` 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`. I0416 20:14:03.374118 132919762885760 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78dcb5361e50>` 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`. I0416 20:14:03.374147 132919762885760 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 0x78dcb5398da0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78dcb5398da0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78dcb5361e50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78dcb5361e50>}). I0416 20:14:03.374495 132919762885760 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 20:14:03.374547 132919762885760 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x78dcb5f79760> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 20:14:03.501536 132919762885760 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints I0416 20:14:03.501757 132919762885760 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_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78dcb6ba2ba0> I0416 20:14:03.501837 132919762885760 checkpointing.py:302] Checkpoint manager created! I0416 20:14:03.580732 132919762885760 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0416 20:14:03.584569 132919762885760 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. I0416 20:14:03.640275 132919762885760 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 I0416 20:14:03.670152 132919762885760 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 20:14:03.670218 132919762885760 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 20:14:04.711004 132919762885760 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 20:14:04.711126 132919762885760 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 20:14:04.819677 4086858 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 I0416 20:14:04.909335 132919762885760 maxtext_utils.py:1805] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0416 20:14:04.909435 132919762885760 maxtext_utils.py:1805] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:14:04.909476 132919762885760 maxtext_utils.py:1805] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0416 20:14:04.909507 132919762885760 maxtext_utils.py:1805] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0416 20:14:04.909533 132919762885760 maxtext_utils.py:1805] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:14:04.909558 132919762885760 maxtext_utils.py:1805] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0416 20:14:04.909584 132919762885760 maxtext_utils.py:1805] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:14:04.909609 132919762885760 maxtext_utils.py:1805] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0416 20:14:04.909640 132919762885760 maxtext_utils.py:1805] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:14:04.909665 132919762885760 maxtext_utils.py:1805] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0416 20:14:04.909689 132919762885760 maxtext_utils.py:1805] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0416 20:14:04.909713 132919762885760 maxtext_utils.py:1805] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0416 20:14:05.026532 132919762885760 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:14:05.026612 132919762885760 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:14:05.031590 132919762885760 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 20:14:05.031641 132919762885760 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:14:05.046914 132919762885760 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:14:05.046964 132919762885760 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:14:05.061547 132919762885760 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 20:14:05.061599 132919762885760 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 20:14:05.083641 132919762885760 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 20:14:05.083696 132919762885760 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:14:05.098402 132919762885760 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 20:14:05.098450 132919762885760 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:14:05.113148 132919762885760 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 20:14:05.113198 132919762885760 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:14:05.141361 132919762885760 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 20:14:05.141424 132919762885760 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 20:14:05.159524 132919762885760 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 20:14:05.159580 132919762885760 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). W0416 20:14:05.519320 4086858 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:14:05.634358 132919762885760 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. I0416 20:14:05.634864 132919762885760 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 20:14:05.636213 132919762885760 metric_logger.py:289] number parameters: 1.104 billion W0416 20:14:08.482226 4086858 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 20:14:08.618680 132919762885760 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 20:14:08.738180 132919762885760 checkpointing.py:798] Waited 0.11948847770690918 seconds for step 0 to finish before starting checkpointing. I0416 20:14:08.738596 132919762885760 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:14:08.738744 132919762885760 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 20:14:08.739125 132919762885760 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0. I0416 20:14:08.841390 132919762885760 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 20:14:08.939458 132919762885760 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:14:08.939548 132919762885760 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 I0416 20:14:08.945422 132806619432512 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 20:14:09.656013 132806608946752 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 20:14:09.831472 132806556517952 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370449568775359, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776370449568775359, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:14:10.246243 4089627 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 20:14:10.246295 4089627 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 20:14:11.079586 132919762885760 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.141422s I0416 20:14:11.086194 132919762885760 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.502 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0416 20:14:11.086282 132919762885760 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.243163s (batch_requests_ready=0.087325s, total_serialization_initiated=2.149386s, others=0.006452s) I0416 20:14:11.086382 132919762885760 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.244008s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=2.243988) I0416 20:14:11.088481 132806634112576 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:14:11.088604 132919762885760 async_checkpointer.py:561] Finished blocking save. Time taken: 2.349813s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0. I0416 20:14:11.088887 132919762885760 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:14:11.089117 132806493603392 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:14:11.089236 132919762885760 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776370448.738584, 'wait_for_prev_duration_secs': 3.910064697265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776370448.7387652, 'checkpointer_blocking_duration_secs': 2.3499999046325684, 'get_old_steps_start_time': 1776370451.088793, 'get_old_steps_duration_secs': 4.076957702636719e-05, 'checkpoint_manager_blocking_start_time': 1776370448.738505, 'checkpoint_manager_blocking_duration_secs': 2.3506908416748047} I0416 20:14:11.089356 132919762885760 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 20:14:11.089427 132919762885760 max_utils.py:750] Memstats: After params initialized: I0416 20:14:11.089480 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 20:14:11.089509 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 20:14:11.089529 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 20:14:11.089548 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 20:14:11.089567 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 20:14:11.089586 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 20:14:11.089603 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 20:14:11.089620 132919762885760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0416 20:14:11.521457 132919762885760 metric_logger.py:185] completed step: 0, seconds: 2.981, TFLOP/s/device: 4.558, Tokens/s/device: 686.961, total_weights: 13328, loss: 10.880 I0416 20:14:11.522345 132919762885760 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/tensorboard/' I0416 20:14:11.634324 132919762885760 metric_logger.py:185] completed step: 1, seconds: 2.899, TFLOP/s/device: 4.687, Tokens/s/device: 706.430, total_weights: 12332, loss: 10.862 I0416 20:14:11.751430 132919762885760 metric_logger.py:185] completed step: 2, seconds: 0.022, TFLOP/s/device: 611.399, Tokens/s/device: 92156.775, total_weights: 15161, loss: 9.926 I0416 20:14:11.869136 132919762885760 metric_logger.py:185] completed step: 3, seconds: 0.104, TFLOP/s/device: 130.350, Tokens/s/device: 19647.722, total_weights: 13327, loss: 9.396 I0416 20:14:11.895949 132806587975232 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0416 20:14:31.292713 132919762885760 metric_logger.py:185] completed step: 4, seconds: 0.117, TFLOP/s/device: 116.115, Tokens/s/device: 17502.179, total_weights: 11939, loss: 9.005 I0416 20:14:31.304466 132919762885760 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.362, Tokens/s/device: 17388.646, total_weights: 15502, loss: 8.861 I0416 20:14:31.419653 132919762885760 metric_logger.py:185] completed step: 6, seconds: 19.424, TFLOP/s/device: 0.700, Tokens/s/device: 105.438, total_weights: 13864, loss: 8.738 I0416 20:14:31.537513 132919762885760 metric_logger.py:185] completed step: 7, seconds: 0.008, TFLOP/s/device: 1697.542, Tokens/s/device: 255872.064, total_weights: 12988, loss: 8.651 I0416 20:14:31.654999 132919762885760 metric_logger.py:185] completed step: 8, seconds: 0.116, TFLOP/s/device: 117.236, Tokens/s/device: 17671.015, total_weights: 13820, loss: 8.671 I0416 20:14:31.773020 132919762885760 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 20:14:31.774237 132919762885760 checkpointing.py:798] Waited 0.0012323856353759766 seconds for step 9 to finish before starting checkpointing. I0416 20:14:31.774500 132919762885760 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:14:42.519107 132806514574912 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.432123s (commit=30.942146s, array_metadata_write=0.489977s) I0416 20:14:42.520218 132806634112576 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 375.235 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0416 20:14:42.520323 132806634112576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.431659s. I0416 20:14:42.810973 132806634112576 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370449568775359, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:14:43.002267 132806634112576 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 20:14:43.228307 132806556517952 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370449568775359, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 20:14:43.456177 132806634112576 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:14:43.456362 132806634112576 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.596301s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 20:14:43.456983 132806634112576 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 20:14:43.711484 132806634112576 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 20:14:44.451492 132806634112576 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0`. I0416 20:14:44.452112 132806634112576 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.713334s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 20:14:44.452186 132806634112576 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.363523s. I0416 20:14:44.452325 132806493603392 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:14:44.452425 132806493603392 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:14:44.452491 132806493603392 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:14:44.452536 132806493603392 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 20:14:44.452704 132919762885760 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. W0416 20:14:44.452836 132919762885760 checkpoint_manager.py:1441] Waiting for previous save to complete took 12.678342 seconds. If this number is high, consider checkpointing less frequently. I0416 20:14:44.453760 132919762885760 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 20:14:44.454070 132919762885760 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9. I0416 20:14:44.636813 132806493603392 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 20:14:44.653578 132919762885760 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0416 20:14:44.653671 132919762885760 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 I0416 20:14:45.351231 132806483117632 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 20:14:54.753709 132919762885760 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.101070s I0416 20:14:54.761027 132919762885760 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.208 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0416 20:14:54.761102 132919762885760 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.212193s (batch_requests_ready=0.096518s, total_serialization_initiated=10.108446s, others=0.007228s) I0416 20:14:54.761191 132919762885760 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.212710s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=10.212692) I0416 20:14:54.762412 132806514574912 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 20:14:54.762512 132919762885760 async_checkpointer.py:561] Finished blocking save. Time taken: 10.308706s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9. I0416 20:14:54.762699 132919762885760 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 20:14:54.762871 132806688638528 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 20:14:54.762941 132919762885760 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776370471.7744584, 'wait_for_prev_duration_secs': 12.67834210395813, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776370484.4537868, 'checkpointer_blocking_duration_secs': 10.308825492858887, 'get_old_steps_start_time': 1776370494.7626276, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1776370471.7744188, 'checkpoint_manager_blocking_duration_secs': 22.988497734069824} I0416 20:14:54.763084 132919762885760 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 20:14:54.763119 132919762885760 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 20:14:55.535596 132818030036544 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0416 20:15:34.440458 132806709610048 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.678717s (commit=39.206746s, array_metadata_write=0.471971s) I0416 20:15:34.441684 132806514574912 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 253.279 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 49 seconds) (per-host) I0416 20:15:34.441791 132806514574912 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.679237s. I0416 20:15:34.865639 132806514574912 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 20:15:35.299244 132806514574912 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 20:15:35.299429 132806514574912 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.579141s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 20:15:35.300068 132806514574912 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 20:15:35.546192 132806514574912 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 20:15:36.255800 132806514574912 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9`. I0416 20:15:36.256430 132806514574912 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 51.802631s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 20:15:36.256510 132806514574912 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.493957s. I0416 20:15:36.256635 132806688638528 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 20:15:36.256691 132806688638528 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 20:15:36.256758 132806688638528 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 20:15:36.256801 132806688638528 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 20:15:36.256947 132919762885760 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. I0416 20:15:36.257178 132919762885760 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 20:15:36.258012 132919762885760 metric_logger.py:185] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.734, Tokens/s/device: 17293.938, total_weights: 12300, loss: 8.668 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops