2026-04-23 01:53:07.949466: 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 00:00:1776909187.962801 2291978 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776909187.966659 2291978 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776909187.978282 2291978 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909187.978301 2291978 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909187.978303 2291978 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909187.978304 2291978 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 01:53:11.666059: 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 01:53:11.803550 138575182657344 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 01:53:47.721591 138575182657344 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 01:53:47.721771 138575182657344 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 01:53:47.721809 138575182657344 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 01:53:47.721840 138575182657344 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 01:53:47.721953 138575182657344 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 01:53:48.373345 138575182657344 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 01:53:48.373623 138575182657344 checkpointing.py:677] Setting up checkpoint logger... I0423 01:53:48.373671 138575182657344 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 01:53:48.373711 138575182657344 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 01:53:48.374170 138575182657344 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 0x7e0842fbeae0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 01:53:50.731644 138575182657344 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 01:53:50.732178 138575182657344 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 0x7e067334a0f0>}, handler_registry=None I0423 01:53:50.732515 138575182657344 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e067334a0f0>` 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 01:53:50.732562 138575182657344 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e067334a750>` 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 01:53:50.732594 138575182657344 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 0x7e067334a0f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e067334a0f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e067334a750>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e067334a750>}). I0423 01:53:50.732989 138575182657344 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 01:53:50.733081 138575182657344 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7e067336c900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 01:53:50.882191 138575182657344 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints I0423 01:53:50.882451 138575182657344 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_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e0673349310> I0423 01:53:50.882724 138575182657344 checkpointing.py:301] Checkpoint manager created! I0423 01:53:50.982926 138575182657344 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 01:53:50.986554 138575182657344 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 01:53:51.057291 138575182657344 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 01:53:51.084087 138575182657344 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:53:51.084160 138575182657344 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:53:51.701166 138575182657344 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 01:53:51.701271 138575182657344 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:53:52.000707 138575182657344 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:53:52.000799 138575182657344 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:53:52.013492 138575182657344 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:53:52.013543 138575182657344 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:53:52.032893 138575182657344 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:53:52.032955 138575182657344 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:53:52.045747 138575182657344 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:53:52.045800 138575182657344 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:53:52.058489 138575182657344 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:53:52.058539 138575182657344 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:53:52.078613 138575182657344 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 01:53:52.078675 138575182657344 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:53:52.094617 138575182657344 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 01:53:52.094676 138575182657344 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 01:53:52.249784 138575182657344 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 01:53:52.249896 138575182657344 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 01:53:52.824644 138575182657344 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 01:53:52.824759 138575182657344 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 01:53:52.824805 138575182657344 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 01:53:52.824851 138575182657344 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 01:53:52.824892 138575182657344 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 01:53:52.824920 138575182657344 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 01:53:52.824959 138575182657344 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 01:53:52.825006 138575182657344 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 01:53:52.825036 138575182657344 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 01:53:52.825084 138575182657344 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 01:53:52.825122 138575182657344 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 01:53:52.825157 138575182657344 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 01:53:53.532510 138575182657344 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:53:53.532603 138575182657344 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:53:53.544711 138575182657344 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:53:53.544764 138575182657344 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:53:54.330006 138575182657344 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. I0423 01:53:54.330600 138575182657344 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 01:53:54.330752 138575182657344 metric_logger.py:301] number parameters: 1.104 billion I0423 01:53:58.800001 138575182657344 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 01:53:58.921480 138575182657344 checkpointing.py:776] Waited 0.12146544456481934 seconds for step 0 to finish before starting checkpointing. I0423 01:53:58.922242 138575182657344 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:53:58.922396 138575182657344 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 01:53:58.922791 138575182657344 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0. I0423 01:53:59.032561 138575182657344 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 01:53:59.156536 138575182657344 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:53:59.156636 138575182657344 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 01:53:59.169643 138449713628736 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0 I0423 01:54:00.038526 138449703142976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/items I0423 01:54:00.144170 138449426318912 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909239898288938, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776909239898288938, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 01:54:00.545113 2294474 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 01:54:00.545173 2294474 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 01:54:04.000166 138575182657344 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.844174s I0423 01:54:04.001067 138575182657344 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.484 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0423 01:54:04.001132 138575182657344 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.967491s (batch_requests_ready=0.118834s, total_serialization_initiated=4.847882s, others=0.000775s) I0423 01:54:04.001206 138575182657344 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.968081s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=4.968061) I0423 01:54:04.008214 138448759424576 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:54:04.008340 138575182657344 async_checkpointer.py:561] Finished blocking save. Time taken: 5.085903s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0. I0423 01:54:04.008540 138575182657344 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:54:04.008756 138448558097984 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:54:04.008854 138575182657344 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776909238.92223, 'wait_for_prev_duration_secs': 4.363059997558594e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776909238.9224167, 'checkpointer_blocking_duration_secs': 5.086039066314697, 'get_old_steps_start_time': 1776909244.008473, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776909238.9221463, 'checkpoint_manager_blocking_duration_secs': 5.086678981781006} I0423 01:54:04.008976 138575182657344 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 01:54:04.009021 138575182657344 max_utils.py:750] Memstats: After params initialized: I0423 01:54:04.009082 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 01:54:04.009110 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 01:54:04.009130 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 01:54:04.009149 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 01:54:04.009167 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 01:54:04.009188 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 01:54:04.009205 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 01:54:04.009221 138575182657344 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 01:54:04.395057 138575182657344 metric_logger.py:196] completed step: 0, seconds: 4.469, TFLOP/s/device: 3.041, Tokens/s/device: 458.300, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0423 01:54:04.396026 138575182657344 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/tensorboard/' I0423 01:54:04.511502 138575182657344 metric_logger.py:196] completed step: 1, seconds: 5.593, TFLOP/s/device: 2.429, Tokens/s/device: 366.179, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0423 01:54:04.629038 138575182657344 metric_logger.py:196] completed step: 2, seconds: 0.010, TFLOP/s/device: 1333.379, Tokens/s/device: 200981.354, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0423 01:54:04.746612 138575182657344 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.817, Tokens/s/device: 17457.124, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0423 01:54:21.876381 138449476650560 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_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0423 01:54:21.876877 138575182657344 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.429, Tokens/s/device: 17398.692, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0423 01:54:21.887456 138575182657344 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 115.535, Tokens/s/device: 17414.670, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0423 01:54:22.004516 138575182657344 metric_logger.py:196] completed step: 6, seconds: 17.131, TFLOP/s/device: 0.793, Tokens/s/device: 119.552, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0423 01:54:22.122471 138575182657344 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1978.036, Tokens/s/device: 298151.114, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0423 01:54:22.240225 138575182657344 metric_logger.py:196] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.241, Tokens/s/device: 17370.358, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0423 01:54:22.357053 138575182657344 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 01:54:22.357890 138575182657344 checkpointing.py:776] Waited 0.0008702278137207031 seconds for step 9 to finish before starting checkpointing. I0423 01:54:22.358208 138575182657344 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:54:37.952171 138448769910336 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.944671s (commit=33.457741s, array_metadata_write=0.486929s) I0423 01:54:37.953449 138448759424576 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 324.688 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0423 01:54:37.953573 138448759424576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.945186s. I0423 01:54:38.239520 138448759424576 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909239898288938, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 01:54:38.461023 138448759424576 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 01:54:38.647945 138449426318912 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909239898288938, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0423 01:54:38.820186 138448759424576 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:54:38.820417 138448759424576 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.534099s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/items I0423 01:54:38.821304 138448759424576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0/items I0423 01:54:39.041234 138448759424576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0 I0423 01:54:39.748123 138448759424576 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_10_shardy_false/checkpoints/0`. I0423 01:54:39.748890 138448759424576 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.826461s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/0 I0423 01:54:39.748970 138448759424576 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.740585s. I0423 01:54:39.749180 138448558097984 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:54:39.749297 138448558097984 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:54:39.749353 138448558097984 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:54:39.749398 138448558097984 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 01:54:39.749526 138575182657344 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 01:54:39.749662 138575182657344 checkpoint_manager.py:1441] Waiting for previous save to complete took 17.391461 seconds. If this number is high, consider checkpointing less frequently. I0423 01:54:39.750813 138575182657344 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 01:54:39.751139 138575182657344 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9. I0423 01:54:39.910362 138448759424576 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9 I0423 01:54:39.920574 138575182657344 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:54:39.920678 138575182657344 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 01:54:40.589799 138449476650560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9/items I0423 01:54:44.037011 138575182657344 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.116860s I0423 01:54:44.041383 138575182657344 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.934 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0423 01:54:44.041450 138575182657344 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.206841s (batch_requests_ready=0.082583s, total_serialization_initiated=4.120015s, others=0.004243s) I0423 01:54:44.041528 138575182657344 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.207430s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=4.207411) I0423 01:54:44.042981 138448759424576 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:54:44.043081 138575182657344 async_checkpointer.py:561] Finished blocking save. Time taken: 4.292224s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9. I0423 01:54:44.043277 138575182657344 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:54:44.043468 138448558097984 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:54:44.043611 138575182657344 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776909262.3581698, 'wait_for_prev_duration_secs': 17.39146113395691, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776909279.750838, 'checkpointer_blocking_duration_secs': 4.292351007461548, 'get_old_steps_start_time': 1776909284.0432067, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776909262.3581266, 'checkpoint_manager_blocking_duration_secs': 21.685460090637207} I0423 01:54:44.043693 138575182657344 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 01:54:44.043725 138575182657344 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:54:44.811555 138558679549504 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_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0423 01:55:21.707942 138449476650560 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.665796s (commit=37.213809s, array_metadata_write=0.451986s) I0423 01:55:21.708991 138448759424576 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 301.779 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0423 01:55:21.709061 138448759424576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.665920s. I0423 01:55:22.179332 138448759424576 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 01:55:22.552520 138448759424576 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:55:22.552706 138448759424576 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557201s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9/items I0423 01:55:22.553438 138448759424576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9/items I0423 01:55:22.788113 138448759424576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9 I0423 01:55:23.475018 138448759424576 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_10_shardy_false/checkpoints/9`. I0423 01:55:23.475746 138448759424576 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.724895s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_10_shardy_false/checkpoints/9 I0423 01:55:23.475827 138448759424576 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.432702s. I0423 01:55:23.475991 138448558097984 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:55:23.476107 138448558097984 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:55:23.476168 138448558097984 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:55:23.476211 138448558097984 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 01:55:23.480013 138575182657344 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 01:55:23.480171 138575182657344 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:55:23.481027 138575182657344 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.233, Tokens/s/device: 17369.180, total_weights: 12300, loss: 8.710, lm_loss: 8.710, perplexity: 6065.008 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops