2026-04-23 01:27:26.096220: 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 00:00:1776907646.109176 2262575 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776907646.112977 2262575 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776907646.124463 2262575 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907646.124477 2262575 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907646.124479 2262575 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907646.124481 2262575 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'} 2026-04-23 01:27:29.788008: 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) I0423 01:27:29.923590 133238938142528 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 01:28:00.258003 133238938142528 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 01:28:00.258275 133238938142528 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 01:28:00.258324 133238938142528 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 I0423 01:28:00.258356 133238938142528 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 01:28:00.258531 133238938142528 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 01:28:00.911978 133238938142528 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 01:28:00.912259 133238938142528 checkpointing.py:677] Setting up checkpoint logger... I0423 01:28:00.912305 133238938142528 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 01:28:00.912342 133238938142528 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 01:28:00.912761 133238938142528 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 0x792dd49aeed0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 01:28:03.270308 133238938142528 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 01:28:03.270686 133238938142528 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 0x792c755fc3e0>}, handler_registry=None I0423 01:28:03.270915 133238938142528 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x792c755fc3e0>` 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`. I0423 01:28:03.270956 133238938142528 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792bf8f21a00>` 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`. I0423 01:28:03.270986 133238938142528 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 0x792c755fc3e0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x792c755fc3e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792bf8f21a00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x792bf8f21a00>}). I0423 01:28:03.271279 133238938142528 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 01:28:03.271332 133238938142528 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x792bef718720> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 01:28:03.393993 133238938142528 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints I0423 01:28:03.394267 133238938142528 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/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x792bef709be0> I0423 01:28:03.394546 133238938142528 checkpointing.py:301] Checkpoint manager created! I0423 01:28:03.472158 133238938142528 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 01:28:03.475056 133238938142528 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. I0423 01:28:03.535557 133238938142528 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 I0423 01:28:03.557880 133238938142528 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:28:03.557945 133238938142528 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:28:04.171119 133238938142528 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 01:28:04.171222 133238938142528 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0423 01:28:04.474594 133238938142528 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:28:04.474684 133238938142528 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0423 01:28:04.487594 133238938142528 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:28:04.487658 133238938142528 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0423 01:28:04.506974 133238938142528 attentions.py:1154] attentions/query Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:28:04.507033 133238938142528 attentions.py:1154] attentions/query Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0423 01:28:04.520026 133238938142528 attentions.py:1155] attentions/key Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:28:04.520089 133238938142528 attentions.py:1155] attentions/key Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0423 01:28:04.532970 133238938142528 attentions.py:1156] attentions/value Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:28:04.533021 133238938142528 attentions.py:1156] attentions/value Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0423 01:28:04.553250 133238938142528 attentions.py:1198] attentions/out Logical: bfloat16[8,128,16,128]...................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 01:28:04.553316 133238938142528 attentions.py:1198] attentions/out Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0423 01:28:04.575944 133238938142528 linears.py:525] linears/x Logical: bfloat16[8,128,7168]........................................ ('activation_batch', 'activation_length', 'activation_mlp'). I0423 01:28:04.576004 133238938142528 linears.py:525] linears/x Physical: bfloat16[8,128,7168]........................................ ('fsdp', None, None). I0423 01:28:04.738426 133238938142528 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 01:28:04.738533 133238938142528 checkpointing.py:665] 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 I0423 01:28:05.328166 133238938142528 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 01:28:05.328282 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 01:28:05.328325 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0423 01:28:05.328371 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0423 01:28:05.328412 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 01:28:05.328442 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0423 01:28:05.328482 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 01:28:05.328524 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0423 01:28:05.328554 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0423 01:28:05.328582 133238938142528 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0423 01:28:05.328620 133238938142528 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 01:28:05.328655 133238938142528 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 01:28:06.036183 133238938142528 train.py:155] train/xent Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:28:06.036275 133238938142528 train.py:155] train/xent Physical: float32[8,128].............................................. ('fsdp', None). I0423 01:28:06.049311 133238938142528 train.py:162] train/z_loss Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:28:06.049366 133238938142528 train.py:162] train/z_loss Physical: float32[8,128].............................................. ('fsdp', None). I0423 01:28:06.678398 133238938142528 max_utils.py:791] Total memory size: 2.5 GB, Output size: 1.5 GB, Temp size: 1.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0423 01:28:06.678919 133238938142528 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 01:28:06.679088 133238938142528 metric_logger.py:301] number parameters: 1.104 billion I0423 01:28:10.261005 133238938142528 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 01:28:10.327018 133238938142528 checkpointing.py:776] Waited 0.06599998474121094 seconds for step 0 to finish before starting checkpointing. I0423 01:28:10.331497 133238938142528 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:28:10.331654 133238938142528 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 01:28:10.332067 133238938142528 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0. I0423 01:28:10.422252 133238938142528 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 01:28:10.519347 133238938142528 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:28:10.519440 133238938142528 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 I0423 01:28:10.525150 133113216763456 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0 I0423 01:28:11.296405 133113206277696 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/items I0423 01:28:11.456232 133112992368192 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907691192947320, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776907691192947320, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0423 01:28:11.769197 2265155 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 01:28:11.769253 2265155 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 01:28:12.531301 133238938142528 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.012475s I0423 01:28:12.536440 133238938142528 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.841 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 01:28:12.536521 133238938142528 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.112790s (batch_requests_ready=0.091578s, total_serialization_initiated=2.016209s, others=0.005003s) I0423 01:28:12.536622 133238938142528 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.113771s (all_items=0.000021s, per_item={'items': '0.00002122'}, temp_paths=2.113750) I0423 01:28:12.538274 133113227249216 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:28:12.538406 133238938142528 async_checkpointer.py:561] Finished blocking save. Time taken: 2.206705s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0. I0423 01:28:12.538792 133238938142528 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:28:12.539148 133112463885888 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:28:12.539255 133238938142528 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776907690.3314836, 'wait_for_prev_duration_secs': 5.030632019042969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776907690.3316755, 'checkpointer_blocking_duration_secs': 2.2068731784820557, 'get_old_steps_start_time': 1776907692.5385725, 'get_old_steps_duration_secs': 0.0001266002655029297, 'checkpoint_manager_blocking_start_time': 1776907690.3313985, 'checkpoint_manager_blocking_duration_secs': 2.207822322845459} I0423 01:28:12.539394 133238938142528 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 01:28:12.539439 133238938142528 max_utils.py:750] Memstats: After params initialized: I0423 01:28:12.539497 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_0(process=0,(0,0,0,0)) I0423 01:28:12.539527 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_1(process=0,(1,0,0,0)) I0423 01:28:12.539550 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_2(process=0,(0,1,0,0)) I0423 01:28:12.539569 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_3(process=0,(1,1,0,0)) I0423 01:28:12.539588 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_4(process=0,(0,2,0,0)) I0423 01:28:12.539606 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_5(process=0,(1,2,0,0)) I0423 01:28:12.539624 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_6(process=0,(0,3,0,0)) I0423 01:28:12.539642 133238938142528 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_7(process=0,(1,3,0,0)) I0423 01:28:12.913697 133238938142528 metric_logger.py:196] completed step: 0, seconds: 3.581, TFLOP/s/device: 0.224, Tokens/s/device: 35.740, total_weights: 963, loss: 10.834, lm_loss: 10.834, perplexity: 50699.652 I0423 01:28:12.914754 133238938142528 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/tensorboard/' I0423 01:28:12.975528 133238938142528 metric_logger.py:196] completed step: 1, seconds: 2.651, TFLOP/s/device: 0.302, Tokens/s/device: 48.292, total_weights: 945, loss: 10.747, lm_loss: 10.747, perplexity: 46511.984 I0423 01:28:13.039593 133238938142528 metric_logger.py:196] completed step: 2, seconds: 0.011, TFLOP/s/device: 75.505, Tokens/s/device: 12067.503, total_weights: 962, loss: 10.247, lm_loss: 10.247, perplexity: 28185.607 I0423 01:28:13.103773 133238938142528 metric_logger.py:196] completed step: 3, seconds: 0.063, TFLOP/s/device: 12.799, Tokens/s/device: 2045.611, total_weights: 975, loss: 9.829, lm_loss: 9.829, perplexity: 18562.363 I0423 01:28:22.680452 133113046894144 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0423 01:28:22.680761 133238938142528 metric_logger.py:196] completed step: 4, seconds: 0.065, TFLOP/s/device: 12.358, Tokens/s/device: 1975.095, total_weights: 962, loss: 9.393, lm_loss: 9.393, perplexity: 11999.676 I0423 01:28:22.708369 133238938142528 metric_logger.py:196] completed step: 5, seconds: 0.063, TFLOP/s/device: 12.643, Tokens/s/device: 2020.680, total_weights: 958, loss: 9.249, lm_loss: 9.249, perplexity: 10399.042 I0423 01:28:22.771570 133238938142528 metric_logger.py:196] completed step: 6, seconds: 9.594, TFLOP/s/device: 0.083, Tokens/s/device: 13.342, total_weights: 930, loss: 9.424, lm_loss: 9.424, perplexity: 12377.827 I0423 01:28:22.836227 133238938142528 metric_logger.py:196] completed step: 7, seconds: 0.008, TFLOP/s/device: 96.783, Tokens/s/device: 15468.278, total_weights: 935, loss: 9.317, lm_loss: 9.317, perplexity: 11127.848 I0423 01:28:22.903536 133238938142528 metric_logger.py:196] completed step: 8, seconds: 0.066, TFLOP/s/device: 12.222, Tokens/s/device: 1953.334, total_weights: 930, loss: 9.479, lm_loss: 9.479, perplexity: 13077.248 I0423 01:28:22.971541 133238938142528 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 01:28:22.972791 133238938142528 checkpointing.py:776] Waited 0.0012698173522949219 seconds for step 9 to finish before starting checkpointing. I0423 01:28:22.973155 133238938142528 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:28:45.230241 133112598103616 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.692880s (commit=32.227696s, array_metadata_write=0.465183s) I0423 01:28:45.231379 133113227249216 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 363.047 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0423 01:28:45.231504 133113227249216 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.693033s. I0423 01:28:45.504996 133113227249216 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907691192947320, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0423 01:28:45.716781 133113227249216 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 01:28:45.916239 133112992368192 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907691192947320, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0423 01:28:46.094002 133113227249216 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:28:46.094282 133113227249216 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.548079s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/items I0423 01:28:46.095083 133113227249216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0/items I0423 01:28:46.327185 133113227249216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0 I0423 01:28:47.018611 133113227249216 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0`. I0423 01:28:47.019371 133113227249216 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.687680s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/0 I0423 01:28:47.019458 133113227249216 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.480993s. I0423 01:28:47.019680 133112463885888 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:28:47.019807 133112463885888 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:28:47.019879 133112463885888 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:28:47.019930 133112463885888 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 01:28:47.020083 133238938142528 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. W0423 01:28:47.020235 133238938142528 checkpoint_manager.py:1441] Waiting for previous save to complete took 24.047094 seconds. If this number is high, consider checkpointing less frequently. I0423 01:28:47.021552 133238938142528 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 01:28:47.021909 133238938142528 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9. I0423 01:28:47.198205 133113227249216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9 I0423 01:28:47.198525 133238938142528 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:28:47.198582 133238938142528 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 I0423 01:28:47.870207 133098767386176 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9/items I0423 01:28:52.259659 133238938142528 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.061751s I0423 01:28:52.264177 133238938142528 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.391 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0423 01:28:52.264244 133238938142528 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.160915s (batch_requests_ready=0.091412s, total_serialization_initiated=5.065101s, others=0.004403s) I0423 01:28:52.264407 133238938142528 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.161608s (all_items=0.000028s, per_item={'items': '0.00002813'}, temp_paths=5.161580) I0423 01:28:52.266179 133112463885888 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:28:52.266264 133238938142528 async_checkpointer.py:561] Finished blocking save. Time taken: 5.244665s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9. I0423 01:28:52.266490 133238938142528 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:28:52.266690 133101730661952 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:28:52.266817 133238938142528 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776907702.973103, 'wait_for_prev_duration_secs': 24.047093868255615, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776907727.0215771, 'checkpointer_blocking_duration_secs': 5.244792222976685, 'get_old_steps_start_time': 1776907732.2663872, 'get_old_steps_duration_secs': 6.318092346191406e-05, 'checkpoint_manager_blocking_start_time': 1776907702.9730396, 'checkpoint_manager_blocking_duration_secs': 29.29375147819519} I0423 01:28:52.266929 133238938142528 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 01:28:52.266963 133238938142528 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:28:52.975386 133222165907008 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0423 01:29:31.829413 133107575424576 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.564686s (commit=39.092842s, array_metadata_write=0.471844s) I0423 01:29:31.830594 133112463885888 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 282.530 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0423 01:29:31.830650 133112463885888 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.564342s. I0423 01:29:32.295887 133112463885888 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 01:29:32.677115 133112463885888 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:29:32.683375 133112463885888 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557101s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9/items I0423 01:29:32.684157 133112463885888 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9/items I0423 01:29:32.914032 133112463885888 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9 I0423 01:29:33.574934 133112463885888 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9`. I0423 01:29:33.575670 133112463885888 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.554076s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_03_dropout/checkpoints/9 I0423 01:29:33.575755 133112463885888 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.309447s. I0423 01:29:33.575951 133101730661952 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:29:33.576078 133101730661952 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:29:33.576149 133101730661952 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:29:33.576193 133101730661952 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 01:29:33.576376 133238938142528 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. I0423 01:29:33.576544 133238938142528 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:29:33.582769 133238938142528 metric_logger.py:196] completed step: 9, seconds: 0.065, TFLOP/s/device: 12.386, Tokens/s/device: 1979.647, total_weights: 948, loss: 9.141, lm_loss: 9.141, perplexity: 9328.599 Per train step: Total TFLOPs: 0.80 split as 99.60% learnable weight flops and 0.40% attention flops