2026-04-22 17:50:44.552927: 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 00:00:1776880244.566072 2021334 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776880244.569930 2021334 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776880244.581412 2021334 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880244.581425 2021334 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880244.581426 2021334 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880244.581428 2021334 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'} 2026-04-22 17:50:48.265156: 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) I0422 17:50:48.404293 140084327577408 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 17:51:17.442080 140084327577408 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 17:51:17.442355 140084327577408 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 17:51:17.442403 140084327577408 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 I0422 17:51:17.442435 140084327577408 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 17:51:17.442622 140084327577408 train.py:706] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 17:51:18.097209 140084327577408 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 17:51:18.097476 140084327577408 checkpointing.py:677] Setting up checkpoint logger... I0422 17:51:18.097520 140084327577408 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 17:51:18.097557 140084327577408 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 17:51:18.097999 140084327577408 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 0x7f67a59a2f90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 17:51:20.553452 140084327577408 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 17:51:20.553844 140084327577408 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 0x7f65cb369d00>}, handler_registry=None I0422 17:51:20.554091 140084327577408 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f65cb369d00>` 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`. I0422 17:51:20.554133 140084327577408 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f65cb36a270>` 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`. I0422 17:51:20.554162 140084327577408 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 0x7f65cb369d00>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f65cb369d00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f65cb36a270>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f65cb36a270>}). I0422 17:51:20.554428 140084327577408 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 17:51:20.554479 140084327577408 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7f65cb37c720> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 17:51:20.711994 140084327577408 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints I0422 17:51:20.712365 140084327577408 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_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f65cb368c80> I0422 17:51:20.712660 140084327577408 checkpointing.py:301] Checkpoint manager created! I0422 17:51:20.793273 140084327577408 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0422 17:51:20.796460 140084327577408 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. I0422 17:51:20.864305 140084327577408 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 I0422 17:51:20.890184 140084327577408 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0422 17:51:20.890257 140084327577408 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0422 17:51:21.509350 140084327577408 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 17:51:21.509450 140084327577408 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0422 17:51:21.812800 140084327577408 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:51:21.812889 140084327577408 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0422 17:51:21.825675 140084327577408 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:51:21.825732 140084327577408 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0422 17:51:21.844815 140084327577408 attentions.py:1154] attentions/query Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:51:21.844876 140084327577408 attentions.py:1154] attentions/query Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0422 17:51:21.857751 140084327577408 attentions.py:1155] attentions/key Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:51:21.857803 140084327577408 attentions.py:1155] attentions/key Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0422 17:51:21.870580 140084327577408 attentions.py:1156] attentions/value Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:51:21.870634 140084327577408 attentions.py:1156] attentions/value Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0422 17:51:21.890672 140084327577408 attentions.py:1198] attentions/out Logical: bfloat16[8,128,16,128]...................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 17:51:21.890739 140084327577408 attentions.py:1198] attentions/out Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0422 17:51:21.913286 140084327577408 linears.py:525] linears/x Logical: bfloat16[8,128,7168]........................................ ('activation_batch', 'activation_length', 'activation_mlp'). I0422 17:51:21.913343 140084327577408 linears.py:525] linears/x Physical: bfloat16[8,128,7168]........................................ ('fsdp', None, None). I0422 17:51:22.076107 140084327577408 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 17:51:22.076208 140084327577408 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 I0422 17:51:22.663779 140084327577408 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 17:51:22.663895 140084327577408 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) I0422 17:51:22.663937 140084327577408 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) I0422 17:51:22.663987 140084327577408 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') I0422 17:51:22.664028 140084327577408 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) I0422 17:51:22.664071 140084327577408 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) I0422 17:51:22.664118 140084327577408 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) I0422 17:51:22.664157 140084327577408 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') I0422 17:51:22.664187 140084327577408 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) I0422 17:51:22.664214 140084327577408 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) I0422 17:51:22.664247 140084327577408 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 17:51:22.664282 140084327577408 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 17:51:23.394503 140084327577408 train.py:155] train/xent Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:51:23.394598 140084327577408 train.py:155] train/xent Physical: float32[8,128].............................................. ('fsdp', None). I0422 17:51:23.406778 140084327577408 train.py:162] train/z_loss Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:51:23.406830 140084327577408 train.py:162] train/z_loss Physical: float32[8,128].............................................. ('fsdp', None). I0422 17:51:27.383246 140084327577408 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. I0422 17:51:27.383915 140084327577408 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 17:51:27.384081 140084327577408 metric_logger.py:301] number parameters: 1.104 billion I0422 17:51:33.745608 140084327577408 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 17:51:33.818497 140084327577408 checkpointing.py:776] Waited 0.07291650772094727 seconds for step 0 to finish before starting checkpointing. I0422 17:51:33.823539 140084327577408 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:51:33.823752 140084327577408 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 17:51:33.824490 140084327577408 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0. I0422 17:51:33.932753 140084327577408 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 17:51:34.041669 139956437648960 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0 I0422 17:51:34.056179 140084327577408 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 17:51:34.056274 140084327577408 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 I0422 17:51:34.834500 139956427163200 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/items I0422 17:51:34.984536 139956169213504 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880294730025005, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776880294730025005, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0422 17:51:35.307304 2024498 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 17:51:35.307359 2024498 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 17:51:35.554626 140084327577408 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.498955s I0422 17:51:35.559437 140084327577408 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.594 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0422 17:51:35.559731 140084327577408 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.625463s (batch_requests_ready=0.117653s, total_serialization_initiated=1.502885s, others=0.004925s) I0422 17:51:35.559818 140084327577408 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.626492s (all_items=0.000040s, per_item={'items': '0.00004005'}, temp_paths=1.626452) I0422 17:51:35.561100 139955621856832 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 17:51:35.561191 140084327577408 async_checkpointer.py:561] Finished blocking save. Time taken: 1.737392s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0. I0422 17:51:35.561397 140084327577408 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:51:35.561578 139956448134720 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:51:35.561674 140084327577408 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880293.8235166, 'wait_for_prev_duration_secs': 5.53131103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880293.8237753, 'checkpointer_blocking_duration_secs': 1.7375288009643555, 'get_old_steps_start_time': 1776880295.561323, 'get_old_steps_duration_secs': 3.719329833984375e-05, 'checkpoint_manager_blocking_start_time': 1776880293.8233318, 'checkpoint_manager_blocking_duration_secs': 1.7383158206939697} I0422 17:51:35.561787 140084327577408 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 17:51:35.561849 140084327577408 max_utils.py:750] Memstats: After params initialized: I0422 17:51:35.561902 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_0(process=0,(0,0,0,0)) I0422 17:51:35.561927 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_1(process=0,(1,0,0,0)) I0422 17:51:35.561945 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_2(process=0,(0,1,0,0)) I0422 17:51:35.561963 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_3(process=0,(1,1,0,0)) I0422 17:51:35.561980 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_4(process=0,(0,2,0,0)) I0422 17:51:35.561996 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_5(process=0,(1,2,0,0)) I0422 17:51:35.562018 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_6(process=0,(0,3,0,0)) I0422 17:51:35.562034 140084327577408 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_7(process=0,(1,3,0,0)) I0422 17:51:35.933739 140084327577408 metric_logger.py:196] completed step: 0, seconds: 6.361, TFLOP/s/device: 0.126, Tokens/s/device: 20.123, total_weights: 963, loss: 10.834, lm_loss: 10.834, perplexity: 50699.652 I0422 17:51:35.934737 140084327577408 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/tensorboard/' I0422 17:51:35.995407 140084327577408 metric_logger.py:196] completed step: 1, seconds: 2.186, TFLOP/s/device: 0.366, Tokens/s/device: 58.557, total_weights: 945, loss: 10.747, lm_loss: 10.747, perplexity: 46511.984 I0422 17:51:36.060262 140084327577408 metric_logger.py:196] completed step: 2, seconds: 0.011, TFLOP/s/device: 72.106, Tokens/s/device: 11524.264, total_weights: 962, loss: 10.247, lm_loss: 10.247, perplexity: 28185.607 I0422 17:51:36.124408 140084327577408 metric_logger.py:196] completed step: 3, seconds: 0.063, TFLOP/s/device: 12.722, Tokens/s/device: 2033.263, total_weights: 975, loss: 9.829, lm_loss: 9.829, perplexity: 18562.363 I0422 17:51:45.713863 140084327577408 metric_logger.py:196] completed step: 4, seconds: 0.066, TFLOP/s/device: 12.199, Tokens/s/device: 1949.704, total_weights: 962, loss: 9.393, lm_loss: 9.393, perplexity: 11999.676 I0422 17:51:45.716135 139955760268864 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0422 17:51:45.739258 140084327577408 metric_logger.py:196] completed step: 5, seconds: 0.063, TFLOP/s/device: 12.634, Tokens/s/device: 2019.278, total_weights: 958, loss: 9.249, lm_loss: 9.249, perplexity: 10399.042 I0422 17:51:45.804008 140084327577408 metric_logger.py:196] completed step: 6, seconds: 9.603, TFLOP/s/device: 0.083, Tokens/s/device: 13.329, total_weights: 930, loss: 9.424, lm_loss: 9.424, perplexity: 12377.827 I0422 17:51:45.873208 140084327577408 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 92.235, Tokens/s/device: 14741.449, total_weights: 935, loss: 9.317, lm_loss: 9.317, perplexity: 11127.848 I0422 17:51:45.941265 140084327577408 metric_logger.py:196] completed step: 8, seconds: 0.066, TFLOP/s/device: 12.138, Tokens/s/device: 1939.952, total_weights: 930, loss: 9.479, lm_loss: 9.479, perplexity: 13077.248 I0422 17:51:46.005576 140084327577408 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0422 17:51:46.006942 140084327577408 checkpointing.py:776] Waited 0.0013895034790039062 seconds for step 9 to finish before starting checkpointing. I0422 17:51:46.007310 140084327577408 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:52:16.820555 139955749783104 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.260215s (commit=40.778161s, array_metadata_write=0.482054s) I0422 17:52:16.821736 139955621856832 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 294.650 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 42 seconds) (per-host) I0422 17:52:16.821794 139955621856832 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.260558s. I0422 17:52:17.087466 139955621856832 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880294730025005, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0422 17:52:17.300451 139955621856832 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 17:52:17.526101 139956169213504 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880294730025005, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0422 17:52:17.700689 139955621856832 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 17:52:17.700947 139955621856832 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570277s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/items I0422 17:52:17.701802 139955621856832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0/items I0422 17:52:17.948368 139955621856832 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0 I0422 17:52:18.668075 139955621856832 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0`. I0422 17:52:18.668813 139955621856832 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 44.845020s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/0 I0422 17:52:18.668899 139955621856832 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.107661s. I0422 17:52:18.669095 139956448134720 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:52:18.669210 139956448134720 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:52:18.669281 139956448134720 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:52:18.669329 139956448134720 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 17:52:18.673105 140084327577408 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. W0422 17:52:18.673243 140084327577408 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.665946 seconds. If this number is high, consider checkpointing less frequently. I0422 17:52:18.674374 140084327577408 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0422 17:52:18.674678 140084327577408 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9. I0422 17:52:18.847518 139955621856832 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9 I0422 17:52:18.855128 140084327577408 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 17:52:18.855234 140084327577408 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 I0422 17:52:19.564646 139955760268864 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9/items I0422 17:52:22.596448 140084327577408 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.741803s I0422 17:52:22.600752 140084327577408 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.218 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 17:52:22.600896 140084327577408 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.835188s (batch_requests_ready=0.084887s, total_serialization_initiated=3.746044s, others=0.004257s) I0422 17:52:22.600974 140084327577408 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.835764s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=3.835747) I0422 17:52:22.602728 139955749783104 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 17:52:22.602794 140084327577408 async_checkpointer.py:561] Finished blocking save. Time taken: 3.928356s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9. I0422 17:52:22.603020 140084327577408 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:52:22.603230 139956448134720 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:52:22.603331 140084327577408 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880306.0072732, 'wait_for_prev_duration_secs': 32.66594576835632, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880338.6744046, 'checkpointer_blocking_duration_secs': 3.9285225868225098, 'get_old_steps_start_time': 1776880342.6029463, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776880306.0072224, 'checkpoint_manager_blocking_duration_secs': 36.59608197212219} I0422 17:52:22.603462 140084327577408 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0422 17:52:22.603496 140084327577408 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:52:23.349657 139956328597056 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0422 17:52:59.923808 139956034995776 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.322354s (commit=36.860804s, array_metadata_write=0.461550s) I0422 17:52:59.924914 139955749783104 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 307.022 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0422 17:52:59.925022 139955749783104 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.322173s. I0422 17:53:00.422235 139955749783104 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 17:53:00.854426 139955749783104 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 17:53:00.854610 139955749783104 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.595650s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9/items I0422 17:53:00.855446 139955749783104 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9/items I0422 17:53:01.109078 139955749783104 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9 I0422 17:53:01.835316 139955749783104 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9`. I0422 17:53:01.836073 139955749783104 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.161629s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_03_dropout/checkpoints/9 I0422 17:53:01.836154 139955749783104 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.233307s. I0422 17:53:01.836368 139956448134720 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:53:01.836489 139956448134720 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:53:01.836540 139956448134720 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:53:01.836579 139956448134720 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 17:53:01.836715 140084327577408 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. I0422 17:53:01.836857 140084327577408 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:53:01.837818 140084327577408 metric_logger.py:196] completed step: 9, seconds: 0.070, TFLOP/s/device: 11.474, Tokens/s/device: 1833.837, 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