2026-04-21 18:14:12.660829: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776795252.671538 1702406 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776795252.674793 1702406 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776795252.685749 1702406 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776795252.685761 1702406 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776795252.685763 1702406 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776795252.685764 1702406 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-21 18:14:16.326527: 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) I0421 18:14:16.461271 127245544942720 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 18:15:13.679276 127245544942720 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 18:15:13.679528 127245544942720 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 18:15:13.679576 127245544942720 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0421 18:15:13.679605 127245544942720 train_utils.py:378] WARNING: 'dataset_path' might be pointing your local file system I0421 18:15:13.679786 127245544942720 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 18:15:14.312823 127245544942720 maxtext_utils.py:1732] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 18:15:14.410619 127245544942720 checkpointing.py:688] Setting up checkpoint logger... I0421 18:15:14.410723 127245544942720 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 18:15:14.410763 127245544942720 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 18:15:14.410938 127245544942720 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 0x73ba5fdab2c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 18:15:16.883763 127245544942720 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0421 18:15:16.883935 127245544942720 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 0x73b908d2d2b0>}, handler_registry=None I0421 18:15:16.884171 127245544942720 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x73b908d2d2b0>` 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`. I0421 18:15:16.884212 127245544942720 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x73b8b63f59a0>` 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`. I0421 18:15:16.884241 127245544942720 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 0x73b908d2d2b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x73b908d2d2b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x73b8b63f59a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x73b8b63f59a0>}). I0421 18:15:16.884527 127245544942720 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 18:15:16.884580 127245544942720 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x73b8b7ba8c20> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 18:15:17.018911 127245544942720 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints I0421 18:15:17.019162 127245544942720 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_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x73b8b63f7050> I0421 18:15:17.019239 127245544942720 checkpointing.py:302] Checkpoint manager created! I0421 18:15:17.095017 127245544942720 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0421 18:15:17.097819 127245544942720 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. I0421 18:15:17.150978 127245544942720 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 I0421 18:15:17.173966 127245544942720 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0421 18:15:17.174029 127245544942720 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0421 18:15:17.813405 127245544942720 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0421 18:15:17.813512 127245544942720 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 I0421 18:15:20.477013 127245544942720 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 18:15:20.477243 127245544942720 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 18:15:20.482620 127245544942720 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 18:15:20.482772 127245544942720 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 18:15:20.496102 127245544942720 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 18:15:20.496184 127245544942720 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 18:15:20.508649 127245544942720 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 18:15:20.508724 127245544942720 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 18:15:20.529202 127245544942720 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 18:15:20.529288 127245544942720 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 18:15:20.541967 127245544942720 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 18:15:20.542059 127245544942720 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 18:15:20.554963 127245544942720 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 18:15:20.555059 127245544942720 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 18:15:20.577834 127245544942720 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 18:15:20.577931 127245544942720 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 18:15:20.595668 127245544942720 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 18:15:20.595753 127245544942720 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 18:15:28.405485 127245544942720 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. I0421 18:15:28.409118 127245544942720 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 18:15:28.410619 127245544942720 metric_logger.py:301] number parameters: 1.104 billion I0421 18:15:40.572052 127245544942720 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0421 18:15:40.703060 127245544942720 checkpointing.py:798] Waited 0.13103270530700684 seconds for step 0 to finish before starting checkpointing. I0421 18:15:40.703725 127245544942720 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 18:15:40.703926 127245544942720 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0421 18:15:40.704555 127245544942720 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0. I0421 18:15:40.810654 127245544942720 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 18:15:40.923672 127153966417472 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0 I0421 18:15:40.946284 127245544942720 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0421 18:15:40.946374 127245544942720 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 I0421 18:15:41.643187 127153955931712 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/items I0421 18:15:41.780808 127153924474432 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776795341550729609, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776795341550729609, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0421 18:15:42.178178 1706358 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 18:15:42.178238 1706358 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 18:15:43.473129 127245544942720 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.528125s I0421 18:15:43.479771 127245544942720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.626 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0421 18:15:43.479967 127245544942720 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.667702s (batch_requests_ready=0.124008s, total_serialization_initiated=2.537022s, others=0.006672s) I0421 18:15:43.480062 127245544942720 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.668891s (all_items=0.000036s, per_item={'items': '0.00003624'}, temp_paths=2.668855) I0421 18:15:43.484439 127153851074112 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 18:15:43.484520 127245544942720 async_checkpointer.py:561] Finished blocking save. Time taken: 2.780547s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0. I0421 18:15:43.484729 127245544942720 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 18:15:43.484902 127153641358912 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 18:15:43.484987 127245544942720 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776795340.7037096, 'wait_for_prev_duration_secs': 4.458427429199219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776795340.7039485, 'checkpointer_blocking_duration_secs': 2.780679702758789, 'get_old_steps_start_time': 1776795343.484648, 'get_old_steps_duration_secs': 3.814697265625e-05, 'checkpoint_manager_blocking_start_time': 1776795340.703543, 'checkpoint_manager_blocking_duration_secs': 2.781419515609741} I0421 18:15:43.485096 127245544942720 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0421 18:15:43.485157 127245544942720 max_utils.py:750] Memstats: After params initialized: I0421 18:15:43.485208 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 18:15:43.485234 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 18:15:43.485254 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 18:15:43.485274 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 18:15:43.485291 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 18:15:43.485309 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 18:15:43.485326 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 18:15:43.485342 127245544942720 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 18:15:43.916980 127245544942720 metric_logger.py:196] completed step: 0, seconds: 12.160, TFLOP/s/device: 1.117, Tokens/s/device: 168.422, total_weights: 13328, loss: 10.880, lm_loss: 10.880, perplexity: 53108.402 I0421 18:15:43.917890 127245544942720 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/tensorboard/' I0421 18:15:44.029240 127245544942720 metric_logger.py:196] completed step: 1, seconds: 3.341, TFLOP/s/device: 4.067, Tokens/s/device: 613.075, total_weights: 12332, loss: 10.862, lm_loss: 10.862, perplexity: 52152.828 I0421 18:15:44.146844 127245544942720 metric_logger.py:196] completed step: 2, seconds: 0.022, TFLOP/s/device: 607.518, Tokens/s/device: 91571.652, total_weights: 15161, loss: 9.926, lm_loss: 9.926, perplexity: 20464.070 I0421 18:15:44.255226 127153934960192 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_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0421 18:15:44.264763 127245544942720 metric_logger.py:196] completed step: 3, seconds: 0.105, TFLOP/s/device: 129.739, Tokens/s/device: 19555.606, total_weights: 13327, loss: 9.396, lm_loss: 9.396, perplexity: 12037.986 I0421 18:16:00.835080 127245544942720 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.405, Tokens/s/device: 17395.145, total_weights: 11939, loss: 9.005, lm_loss: 9.005, perplexity: 8141.998 I0421 18:16:00.847563 127245544942720 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.328, Tokens/s/device: 17383.481, total_weights: 15502, loss: 8.861, lm_loss: 8.861, perplexity: 7054.059 I0421 18:16:00.962419 127245544942720 metric_logger.py:196] completed step: 6, seconds: 16.570, TFLOP/s/device: 0.820, Tokens/s/device: 123.595, total_weights: 13864, loss: 8.738, lm_loss: 8.738, perplexity: 6233.198 I0421 18:16:01.080296 127245544942720 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1555.303, Tokens/s/device: 234432.234, total_weights: 12988, loss: 8.651, lm_loss: 8.651, perplexity: 5718.487 I0421 18:16:01.198262 127245544942720 metric_logger.py:196] completed step: 8, seconds: 0.116, TFLOP/s/device: 117.246, Tokens/s/device: 17672.540, total_weights: 13820, loss: 8.671, lm_loss: 8.671, perplexity: 5832.209 I0421 18:16:01.316005 127245544942720 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0421 18:16:01.317230 127245544942720 checkpointing.py:798] Waited 0.0012476444244384766 seconds for step 9 to finish before starting checkpointing. I0421 18:16:01.317534 127245544942720 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 18:16:15.091368 127153872045632 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.610910s (commit=31.138888s, array_metadata_write=0.472022s) I0421 18:16:15.092547 127153851074112 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 368.632 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0421 18:16:15.092659 127153851074112 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.608089s. I0421 18:16:15.321366 127153851074112 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776795341550729609, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0421 18:16:15.515677 127153851074112 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 18:16:15.743094 127153924474432 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776795341550729609, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0421 18:16:15.921312 127153851074112 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 18:16:15.921513 127153851074112 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557210s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/items I0421 18:16:15.922181 127153851074112 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0/items I0421 18:16:16.162638 127153851074112 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0 I0421 18:16:16.835412 127153851074112 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_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0`. I0421 18:16:16.836068 127153851074112 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.132098s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/0 I0421 18:16:16.836149 127153851074112 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.351583s. I0421 18:16:16.836277 127153641358912 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 18:16:16.836331 127153641358912 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 18:16:16.836388 127153641358912 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 18:16:16.836439 127153641358912 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 18:16:16.836559 127245544942720 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. W0421 18:16:16.836685 127245544942720 checkpoint_manager.py:1441] Waiting for previous save to complete took 15.519175 seconds. If this number is high, consider checkpointing less frequently. I0421 18:16:16.837903 127245544942720 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0421 18:16:16.838215 127245544942720 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9. I0421 18:16:17.030067 127245544942720 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array. I0421 18:16:17.030158 127245544942720 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 I0421 18:16:17.037292 127153641358912 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9 I0421 18:16:17.722798 127153840588352 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9/items I0421 18:16:27.978636 127245544942720 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.949842s I0421 18:16:27.985232 127245544942720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.117 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0421 18:16:27.985294 127245544942720 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.052851s (batch_requests_ready=0.088856s, total_serialization_initiated=10.957502s, others=0.006493s) I0421 18:16:27.985380 127245544942720 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.053413s (all_items=0.000023s, per_item={'items': '0.00002265'}, temp_paths=11.053390) I0421 18:16:27.986761 127153893017152 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 18:16:27.986840 127245544942720 async_checkpointer.py:561] Finished blocking save. Time taken: 11.148890s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9. I0421 18:16:27.987073 127245544942720 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 18:16:27.987293 127153872045632 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 18:16:27.987485 127245544942720 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776795361.3174775, 'wait_for_prev_duration_secs': 15.519174575805664, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776795376.8379276, 'checkpointer_blocking_duration_secs': 11.14902925491333, 'get_old_steps_start_time': 1776795387.9869773, 'get_old_steps_duration_secs': 3.9577484130859375e-05, 'checkpoint_manager_blocking_start_time': 1776795361.317421, 'checkpoint_manager_blocking_duration_secs': 26.670037031173706} I0421 18:16:27.987624 127245544942720 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0421 18:16:27.987659 127245544942720 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 18:16:28.749709 127152401942080 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_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0421 18:17:17.287907 127153934960192 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 49.301986s (commit=48.868129s, array_metadata_write=0.433857s) I0421 18:17:17.289131 127153893017152 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 209.368 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host) I0421 18:17:17.289188 127153893017152 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 49.302304s. I0421 18:17:17.735462 127153893017152 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 18:17:18.190894 127153893017152 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 18:17:18.191105 127153893017152 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.596150s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9/items I0421 18:17:18.191726 127153893017152 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9/items I0421 18:17:18.421992 127153893017152 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9 I0421 18:17:19.096149 127153893017152 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_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9`. I0421 18:17:19.096772 127153893017152 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 62.258827s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_10_shardy_false/checkpoints/9 I0421 18:17:19.096849 127153893017152 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 51.109965s. I0421 18:17:19.097069 127153872045632 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 18:17:19.097179 127153872045632 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 18:17:19.097247 127153872045632 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 18:17:19.097296 127153872045632 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 18:17:19.097458 127245544942720 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. I0421 18:17:19.097666 127245544942720 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 18:17:19.099821 127245544942720 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.774, Tokens/s/device: 17299.927, 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