2026-04-23 01:22:49.651023: 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 00:00:1776907369.664084 2256777 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776907369.668021 2256777 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776907369.679184 2256777 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907369.679199 2256777 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907369.679201 2256777 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776907369.679203 2256777 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 01:22:53.338641: 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 01:22:53.472998 129094465443648 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 01:23:25.241438 129094465443648 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 01:23:25.241701 129094465443648 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 01:23:25.241753 129094465443648 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 01:23:25.241786 129094465443648 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 01:23:25.241977 129094465443648 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 01:23:25.898994 129094465443648 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 01:23:25.899276 129094465443648 checkpointing.py:677] Setting up checkpoint logger... I0423 01:23:25.899321 129094465443648 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 01:23:25.899357 129094465443648 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 01:23:25.899802 129094465443648 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 0x7568dc3b3110>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 01:23:28.757752 129094465443648 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 01:23:28.758138 129094465443648 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 0x75678671fe60>}, handler_registry=None I0423 01:23:28.758380 129094465443648 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75678671fe60>` 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 01:23:28.758420 129094465443648 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75677f940410>` 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 01:23:28.758450 129094465443648 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 0x75678671fe60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x75678671fe60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75677f940410>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x75677f940410>}). I0423 01:23:28.758727 129094465443648 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 01:23:28.758780 129094465443648 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x75677ef205e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 01:23:28.880557 129094465443648 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints I0423 01:23:28.880806 129094465443648 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_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x75677ef3aa80> I0423 01:23:28.881093 129094465443648 checkpointing.py:301] Checkpoint manager created! I0423 01:23:28.960963 129094465443648 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 01:23:28.963964 129094465443648 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 01:23:29.024233 129094465443648 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 01:23:29.048067 129094465443648 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:23:29.048133 129094465443648 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:23:29.653299 129094465443648 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 01:23:29.653404 129094465443648 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:23:29.955410 129094465443648 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:23:29.955498 129094465443648 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:23:29.968287 129094465443648 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:23:29.968336 129094465443648 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:23:29.987404 129094465443648 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 01:23:29.987472 129094465443648 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:23:30.000258 129094465443648 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 01:23:30.000311 129094465443648 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:23:30.012895 129094465443648 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 01:23:30.012947 129094465443648 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:23:30.032980 129094465443648 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 01:23:30.033051 129094465443648 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:23:30.048955 129094465443648 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 01:23:30.049017 129094465443648 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 01:23:30.202951 129094465443648 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 01:23:30.203076 129094465443648 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 01:23:30.792784 129094465443648 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 01:23:30.792900 129094465443648 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 01:23:30.792944 129094465443648 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 01:23:30.792998 129094465443648 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 01:23:30.793040 129094465443648 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 01:23:30.793088 129094465443648 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 01:23:30.793130 129094465443648 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 01:23:30.793170 129094465443648 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 01:23:30.793199 129094465443648 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 01:23:30.793227 129094465443648 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 01:23:30.793262 129094465443648 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 01:23:30.793296 129094465443648 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 01:23:31.206777 129094465443648 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:23:31.206871 129094465443648 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:23:31.218844 129094465443648 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:23:31.218894 129094465443648 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:23:32.294873 129094465443648 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 01:23:32.295652 129094465443648 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 01:23:32.295893 129094465443648 metric_logger.py:301] number parameters: 1.104 billion I0423 01:23:36.202038 129094465443648 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 01:23:36.323986 129094465443648 checkpointing.py:776] Waited 0.12196755409240723 seconds for step 0 to finish before starting checkpointing. I0423 01:23:36.324969 129094465443648 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:23:36.325208 129094465443648 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 01:23:36.325869 129094465443648 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0. I0423 01:23:36.417429 129094465443648 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 01:23:36.511772 129094465443648 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:23:36.511876 129094465443648 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 01:23:36.517909 128968873215552 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0 I0423 01:23:37.314846 128968862729792 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/items I0423 01:23:37.449691 128968831272512 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907417211470251, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776907417211470251, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/_CHECKPOINT_METADATA I0423 01:23:37.809411 2259227 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 01:23:37.809480 2259227 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 01:23:38.374248 129094465443648 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.863163s I0423 01:23:38.379319 129094465443648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.294 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 01:23:38.379392 129094465443648 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.960778s (batch_requests_ready=0.088759s, total_serialization_initiated=1.867050s, others=0.004968s) I0423 01:23:38.379467 129094465443648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.961450s (all_items=0.000034s, per_item={'items': '0.00003433'}, temp_paths=1.961415) I0423 01:23:38.380836 128969022113344 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:23:38.380941 129094465443648 async_checkpointer.py:561] Finished blocking save. Time taken: 2.055686s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0. I0423 01:23:38.381159 129094465443648 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:23:38.381363 128967965148736 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:23:38.381490 129094465443648 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776907416.324953, 'wait_for_prev_duration_secs': 4.696846008300781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776907416.3252327, 'checkpointer_blocking_duration_secs': 2.0558276176452637, 'get_old_steps_start_time': 1776907418.3810828, 'get_old_steps_duration_secs': 3.409385681152344e-05, 'checkpoint_manager_blocking_start_time': 1776907416.32479, 'checkpoint_manager_blocking_duration_secs': 2.0566742420196533} I0423 01:23:38.381573 129094465443648 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 01:23:38.381616 129094465443648 max_utils.py:750] Memstats: After params initialized: I0423 01:23:38.381662 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0423 01:23:38.381686 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 01:23:38.381706 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 01:23:38.381730 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 01:23:38.381748 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 01:23:38.381764 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 01:23:38.381781 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 01:23:38.381799 129094465443648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 01:23:38.739205 129094465443648 metric_logger.py:196] completed step: 0, seconds: 3.905, TFLOP/s/device: 3.479, Tokens/s/device: 524.398, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0423 01:23:38.740187 129094465443648 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/tensorboard/' I0423 01:23:38.855752 129094465443648 metric_logger.py:196] completed step: 1, seconds: 2.535, TFLOP/s/device: 5.360, Tokens/s/device: 807.884, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0423 01:23:38.973355 129094465443648 metric_logger.py:196] completed step: 2, seconds: 0.010, TFLOP/s/device: 1321.191, Tokens/s/device: 199144.302, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0423 01:23:39.090781 129094465443648 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.512, Tokens/s/device: 17411.265, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0423 01:23:39.175854 128968841758272 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_01_base/checkpoints/0/items/array_metadatas/process_0 I0423 01:23:56.370507 129094465443648 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.331, Tokens/s/device: 17383.923, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0423 01:23:56.381628 129094465443648 metric_logger.py:196] completed step: 5, seconds: 0.117, TFLOP/s/device: 116.102, Tokens/s/device: 17500.085, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0423 01:23:56.499093 129094465443648 metric_logger.py:196] completed step: 6, seconds: 17.281, TFLOP/s/device: 0.786, Tokens/s/device: 118.514, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0423 01:23:56.616288 129094465443648 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1871.248, Tokens/s/device: 282054.813, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0423 01:23:56.733939 129094465443648 metric_logger.py:196] completed step: 8, seconds: 0.119, TFLOP/s/device: 114.496, Tokens/s/device: 17258.088, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0423 01:23:56.850914 129094465443648 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 01:23:56.851908 129094465443648 checkpointing.py:776] Waited 0.0010266304016113281 seconds for step 9 to finish before starting checkpointing. I0423 01:23:56.852209 129094465443648 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:24:03.029838 128968233584192 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.649680s (commit=24.160506s, array_metadata_write=0.489173s) I0423 01:24:03.030886 128969022113344 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 474.849 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 26 seconds) (per-host) I0423 01:24:03.030940 128969022113344 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.649954s. I0423 01:24:03.282856 128969022113344 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907417211470251, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/_CHECKPOINT_METADATA I0423 01:24:03.501139 128969022113344 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 01:24:03.675845 128968831272512 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776907417211470251, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/_CHECKPOINT_METADATA I0423 01:24:03.884837 128969022113344 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:24:03.885055 128969022113344 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.560180s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/items I0423 01:24:03.886060 128969022113344 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0/items I0423 01:24:04.124096 128969022113344 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0 I0423 01:24:04.808993 128969022113344 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_01_base/checkpoints/0`. I0423 01:24:04.809802 128969022113344 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 28.484554s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/0 I0423 01:24:04.809887 128969022113344 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 26.428902s. I0423 01:24:04.810064 128967965148736 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:24:04.810184 128967965148736 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:24:04.810240 128967965148736 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:24:04.810289 128967965148736 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 01:24:04.815866 129094465443648 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 01:24:04.816011 129094465443648 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.963801 seconds. If this number is high, consider checkpointing less frequently. I0423 01:24:04.816874 129094465443648 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 01:24:04.817200 129094465443648 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9. I0423 01:24:04.987443 128969022113344 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9 I0423 01:24:04.994438 129094465443648 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:24:04.994536 129094465443648 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 01:24:05.712773 128968841758272 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9/items I0423 01:24:05.766924 129094465443648 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.772983s I0423 01:24:05.767383 129094465443648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 14.242 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 866 milliseconds) (per-host) I0423 01:24:05.767435 129094465443648 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.866615s (batch_requests_ready=0.090297s, total_serialization_initiated=0.775986s, others=0.000333s) I0423 01:24:05.767496 129094465443648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.867198s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.867178) I0423 01:24:05.768374 128968636237376 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:24:05.768476 129094465443648 async_checkpointer.py:561] Finished blocking save. Time taken: 0.951558s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9. I0423 01:24:05.768653 129094465443648 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:24:05.768918 129094465443648 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776907436.8521736, 'wait_for_prev_duration_secs': 7.96380090713501, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776907444.8168993, 'checkpointer_blocking_duration_secs': 0.9516777992248535, 'get_old_steps_start_time': 1776907445.7685897, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776907436.85213, 'checkpoint_manager_blocking_duration_secs': 8.916767597198486} I0423 01:24:05.768991 129094465443648 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 01:24:05.769023 129094465443648 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:24:05.857440 128967965148736 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:24:06.885635 129077813970496 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_01_base/checkpoints/9/items/array_metadatas/process_0 I0423 01:24:41.732948 128968233584192 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.566507s (commit=35.098527s, array_metadata_write=0.467980s) I0423 01:24:41.734143 128968636237376 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 343.081 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0423 01:24:41.734260 128968636237376 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.965737s. I0423 01:24:42.259833 128968636237376 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 01:24:42.657785 128968636237376 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:24:42.657974 128968636237376 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.564450s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9/items I0423 01:24:42.658764 128968636237376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9/items I0423 01:24:42.899147 128968636237376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9 I0423 01:24:43.595145 128968636237376 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_01_base/checkpoints/9`. I0423 01:24:43.595873 128968636237376 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.778960s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_01_base/checkpoints/9 I0423 01:24:43.595951 128968636237376 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.827432s. I0423 01:24:43.596160 128967965148736 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:24:43.596271 128967965148736 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:24:43.596328 128967965148736 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:24:43.596370 128967965148736 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 01:24:43.596537 129094465443648 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 01:24:43.596687 129094465443648 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:24:43.597625 129094465443648 metric_logger.py:196] completed step: 9, seconds: 0.117, TFLOP/s/device: 116.079, Tokens/s/device: 17496.647, 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