2026-04-23 01:47:12.660655: 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 00:00:1776908832.673739 2285717 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776908832.677615 2285717 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776908832.689150 2285717 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908832.689166 2285717 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908832.689168 2285717 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908832.689170 2285717 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 01:47:16.359076: 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 01:47:16.494559 131543128213312 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 01:47:47.344728 131543128213312 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 01:47:47.344997 131543128213312 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 01:47:47.345058 131543128213312 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 01:47:47.345089 131543128213312 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 01:47:47.345276 131543128213312 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 01:47:47.998881 131543128213312 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 2, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0423 01:47:47.999162 131543128213312 checkpointing.py:677] Setting up checkpoint logger... I0423 01:47:47.999209 131543128213312 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 01:47:47.999246 131543128213312 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 01:47:47.999669 131543128213312 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 0x77a2fbdb5760>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 01:47:50.501141 131543128213312 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 01:47:50.501513 131543128213312 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 0x77a19e914440>}, handler_registry=None I0423 01:47:50.501760 131543128213312 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x77a19e914440>` 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 01:47:50.501801 131543128213312 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77a11bf42660>` 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 01:47:50.501829 131543128213312 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 0x77a19e914440>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x77a19e914440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77a11bf42660>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x77a11bf42660>}). I0423 01:47:50.502115 131543128213312 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 01:47:50.502167 131543128213312 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x77a11bf647c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 01:47:50.627362 131543128213312 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints I0423 01:47:50.627589 131543128213312 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_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x77a11bf424b0> I0423 01:47:50.627875 131543128213312 checkpointing.py:301] Checkpoint manager created! I0423 01:47:50.706461 131543128213312 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0423 01:47:50.709342 131543128213312 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 01:47:50.769901 131543128213312 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 01:47:50.792428 131543128213312 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:47:50.792491 131543128213312 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0423 01:47:51.405450 131543128213312 nnx_wrappers.py:437] Unknown Logical: bfloat16[2,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 01:47:51.405551 131543128213312 nnx_wrappers.py:437] Unknown Physical: bfloat16[2,2048,2048]....................................... ('fsdp', None, 'tensor'). I0423 01:47:51.705460 131543128213312 attentions.py:1088] attentions/inputs_q Logical: bfloat16[2,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:47:51.705547 131543128213312 attentions.py:1088] attentions/inputs_q Physical: bfloat16[2,2048,2048]....................................... ('fsdp', None, 'tensor'). I0423 01:47:51.718310 131543128213312 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[2,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:47:51.718362 131543128213312 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[2,2048,2048]....................................... ('fsdp', None, 'tensor'). I0423 01:47:51.737559 131543128213312 attentions.py:1154] attentions/query Logical: bfloat16[2,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:47:51.737630 131543128213312 attentions.py:1154] attentions/query Physical: bfloat16[2,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0423 01:47:51.750472 131543128213312 attentions.py:1155] attentions/key Logical: bfloat16[2,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:47:51.750528 131543128213312 attentions.py:1155] attentions/key Physical: bfloat16[2,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0423 01:47:51.763262 131543128213312 attentions.py:1156] attentions/value Logical: bfloat16[2,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:47:51.763310 131543128213312 attentions.py:1156] attentions/value Physical: bfloat16[2,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0423 01:47:51.783260 131543128213312 attentions.py:1198] attentions/out Logical: bfloat16[2,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 01:47:51.783325 131543128213312 attentions.py:1198] attentions/out Physical: bfloat16[2,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0423 01:47:51.799497 131543128213312 linears.py:525] linears/x Logical: bfloat16[2,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 01:47:51.799556 131543128213312 linears.py:525] linears/x Physical: bfloat16[2,2048,7168]....................................... ('fsdp', None, 'tensor'). I0423 01:47:51.955165 131543128213312 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 01:47:51.955265 131543128213312 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: 2 tensor: 4 I0423 01:47:52.575165 131543128213312 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: ('tensor',) I0423 01:47:52.575281 131543128213312 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, 'tensor') I0423 01:47:52.575325 131543128213312 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, 'tensor') I0423 01:47:52.575371 131543128213312 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: ('tensor', None, 'fsdp') I0423 01:47:52.575413 131543128213312 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: ('tensor', None) I0423 01:47:52.575443 131543128213312 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: ('tensor', None) I0423 01:47:52.575484 131543128213312 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, 'tensor', None) I0423 01:47:52.575526 131543128213312 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: ('tensor', None, None, 'fsdp') I0423 01:47:52.575558 131543128213312 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, 'tensor', None) I0423 01:47:52.575587 131543128213312 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, 'tensor', None) I0423 01:47:52.575624 131543128213312 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', 'tensor') I0423 01:47:52.575660 131543128213312 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: ('tensor', 'fsdp') I0423 01:47:53.263406 131543128213312 train.py:155] train/xent Logical: float32[2,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:47:53.263497 131543128213312 train.py:155] train/xent Physical: float32[2,2048]............................................. ('fsdp', None). I0423 01:47:53.276130 131543128213312 train.py:162] train/z_loss Logical: float32[2,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:47:53.276181 131543128213312 train.py:162] train/z_loss Physical: float32[2,2048]............................................. ('fsdp', None). I0423 01:47:53.922442 131543128213312 max_utils.py:791] Total memory size: 2.5 GB, Output size: 1.5 GB, Temp size: 1.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0423 01:47:53.923003 131543128213312 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 01:47:53.923175 131543128213312 metric_logger.py:301] number parameters: 1.104 billion I0423 01:47:58.150603 131543128213312 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 01:47:58.206219 131543128213312 checkpointing.py:776] Waited 0.055606842041015625 seconds for step 0 to finish before starting checkpointing. I0423 01:47:58.210700 131543128213312 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:47:58.210870 131543128213312 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 01:47:58.211283 131543128213312 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0. I0423 01:47:58.297940 131543128213312 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 01:47:58.373681 131417623234112 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0 I0423 01:47:58.384899 131543128213312 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:47:58.384995 131543128213312 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 01:47:59.108579 131417474336320 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/items I0423 01:47:59.259504 131417304467008 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908879017093955, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776908879017093955, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0423 01:47:59.576535 2288204 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 01:47:59.576595 2288204 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 01:48:00.035220 131543128213312 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.650789s I0423 01:48:00.039890 131543128213312 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.090 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 01:48:00.039958 131543128213312 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.740611s (batch_requests_ready=0.081648s, total_serialization_initiated=1.654399s, others=0.004565s) I0423 01:48:00.040040 131543128213312 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.741542s (all_items=0.000021s, per_item={'items': '0.00002098'}, temp_paths=1.741521) I0423 01:48:00.041610 131416834704960 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:48:00.041699 131543128213312 async_checkpointer.py:561] Finished blocking save. Time taken: 1.830788s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0. I0423 01:48:00.041892 131543128213312 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:48:00.042096 131417633719872 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:48:00.042192 131543128213312 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776908878.2106867, 'wait_for_prev_duration_secs': 5.698204040527344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776908878.210892, 'checkpointer_blocking_duration_secs': 1.830918788909912, 'get_old_steps_start_time': 1776908880.0418289, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776908878.2106006, 'checkpoint_manager_blocking_duration_secs': 1.831566333770752} I0423 01:48:00.042302 131543128213312 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 01:48:00.042341 131543128213312 max_utils.py:750] Memstats: After params initialized: I0423 01:48:00.042389 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_0(process=0,(0,0,0,0)) I0423 01:48:00.042414 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_1(process=0,(1,0,0,0)) I0423 01:48:00.042433 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_2(process=0,(0,1,0,0)) I0423 01:48:00.042452 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_3(process=0,(1,1,0,0)) I0423 01:48:00.042470 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_4(process=0,(0,2,0,0)) I0423 01:48:00.042487 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_5(process=0,(1,2,0,0)) I0423 01:48:00.042503 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_6(process=0,(0,3,0,0)) I0423 01:48:00.042520 131543128213312 max_utils.py:756] Using (GB) 1.58 / 31.25 (5.056000%) on TPU_7(process=0,(1,3,0,0)) I0423 01:48:00.384390 131543128213312 metric_logger.py:196] completed step: 0, seconds: 4.227, TFLOP/s/device: 0.804, Tokens/s/device: 121.129, total_weights: 3650, loss: 10.896, lm_loss: 10.896, perplexity: 53934.855 I0423 01:48:00.385436 131543128213312 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/tensorboard/' I0423 01:48:00.435722 131543128213312 metric_logger.py:196] completed step: 1, seconds: 2.232, TFLOP/s/device: 1.522, Tokens/s/device: 229.429, total_weights: 2676, loss: 10.833, lm_loss: 10.833, perplexity: 50674.457 I0423 01:48:00.488192 131543128213312 metric_logger.py:196] completed step: 2, seconds: 0.015, TFLOP/s/device: 224.269, Tokens/s/device: 33804.305, total_weights: 3783, loss: 10.136, lm_loss: 10.136, perplexity: 25223.543 I0423 01:48:00.540878 131543128213312 metric_logger.py:196] completed step: 3, seconds: 0.053, TFLOP/s/device: 64.682, Tokens/s/device: 9749.595, total_weights: 3680, loss: 9.403, lm_loss: 9.403, perplexity: 12120.116 I0423 01:48:09.505797 131543128213312 metric_logger.py:196] completed step: 4, seconds: 0.052, TFLOP/s/device: 65.129, Tokens/s/device: 9816.892, total_weights: 3736, loss: 9.171, lm_loss: 9.171, perplexity: 9614.685 I0423 01:48:09.509527 131417453364800 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_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_0 I0423 01:48:09.539759 131543128213312 metric_logger.py:196] completed step: 5, seconds: 0.052, TFLOP/s/device: 64.789, Tokens/s/device: 9765.774, total_weights: 4096, loss: 9.075, lm_loss: 9.075, perplexity: 8729.922 I0423 01:48:09.592394 131543128213312 metric_logger.py:196] completed step: 6, seconds: 8.984, TFLOP/s/device: 0.378, Tokens/s/device: 56.992, total_weights: 2931, loss: 8.899, lm_loss: 8.899, perplexity: 7322.795 I0423 01:48:09.644480 131543128213312 metric_logger.py:196] completed step: 7, seconds: 0.013, TFLOP/s/device: 253.737, Tokens/s/device: 38246.060, total_weights: 3526, loss: 8.818, lm_loss: 8.818, perplexity: 6751.396 I0423 01:48:09.697278 131543128213312 metric_logger.py:196] completed step: 8, seconds: 0.054, TFLOP/s/device: 62.611, Tokens/s/device: 9437.440, total_weights: 3012, loss: 8.750, lm_loss: 8.750, perplexity: 6308.487 I0423 01:48:09.749245 131543128213312 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 01:48:09.750533 131543128213312 checkpointing.py:776] Waited 0.0013043880462646484 seconds for step 9 to finish before starting checkpointing. I0423 01:48:09.750944 131543128213312 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:48:39.843156 131416968922688 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.802284s (commit=39.340106s, array_metadata_write=0.462178s) I0423 01:48:39.844434 131416834704960 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 304.171 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0423 01:48:39.844567 131416834704960 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.802814s. I0423 01:48:40.097211 131416834704960 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908879017093955, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0423 01:48:40.326643 131416834704960 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 01:48:40.496697 131417304467008 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908879017093955, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0423 01:48:40.703481 131416834704960 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:48:40.703672 131416834704960 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.563635s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/items I0423 01:48:40.704435 131416834704960 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0/items I0423 01:48:40.943346 131416834704960 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0 I0423 01:48:41.652928 131416834704960 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_09_pdb_lt_1/checkpoints/0`. I0423 01:48:41.653651 131416834704960 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.442743s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/0 I0423 01:48:41.653741 131416834704960 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.611995s. I0423 01:48:41.653980 131417633719872 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:48:41.654122 131417633719872 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:48:41.654195 131417633719872 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:48:41.654254 131417633719872 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 01:48:41.654392 131543128213312 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 01:48:41.654533 131543128213312 checkpoint_manager.py:1441] Waiting for previous save to complete took 31.903613 seconds. If this number is high, consider checkpointing less frequently. I0423 01:48:41.655891 131543128213312 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 01:48:41.656250 131543128213312 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9. I0423 01:48:41.832323 131543128213312 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:48:41.832430 131543128213312 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 01:48:41.837536 131416834704960 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9 I0423 01:48:42.538482 131417453364800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9/items I0423 01:48:43.977153 131543128213312 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.145270s I0423 01:48:43.981928 131543128213312 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.531 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0423 01:48:43.982061 131543128213312 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.231417s (batch_requests_ready=0.078249s, total_serialization_initiated=2.148469s, others=0.004699s) I0423 01:48:43.982148 131543128213312 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.232008s (all_items=0.000025s, per_item={'items': '0.00002527'}, temp_paths=2.231983) I0423 01:48:43.983585 131417633719872 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:48:43.983670 131543128213312 async_checkpointer.py:561] Finished blocking save. Time taken: 2.327732s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9. I0423 01:48:43.983886 131543128213312 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:48:43.984095 131416834704960 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:48:43.984214 131543128213312 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776908889.7508838, 'wait_for_prev_duration_secs': 31.903612852096558, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776908921.655918, 'checkpointer_blocking_duration_secs': 2.327861785888672, 'get_old_steps_start_time': 1776908923.9837992, 'get_old_steps_duration_secs': 4.3392181396484375e-05, 'checkpoint_manager_blocking_start_time': 1776908889.7508125, 'checkpoint_manager_blocking_duration_secs': 34.23337650299072} I0423 01:48:43.984331 131543128213312 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 01:48:43.984365 131543128213312 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:48:44.732030 131526350079552 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_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_0 I0423 01:49:23.259945 131417453364800 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.277328s (commit=38.830453s, array_metadata_write=0.446874s) I0423 01:49:23.261054 131417633719872 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 304.425 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0423 01:49:23.261110 131417633719872 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.277396s. I0423 01:49:23.753254 131417633719872 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 01:49:24.105795 131417633719872 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:49:24.106067 131417633719872 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.534041s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9/items I0423 01:49:24.106819 131417633719872 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9/items I0423 01:49:24.356513 131417633719872 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9 I0423 01:49:25.059929 131417633719872 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_09_pdb_lt_1/checkpoints/9`. I0423 01:49:25.060683 131417633719872 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.404751s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_09_pdb_lt_1/checkpoints/9 I0423 01:49:25.060760 131417633719872 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.077047s. I0423 01:49:25.060940 131416834704960 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:49:25.061066 131416834704960 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:49:25.061131 131416834704960 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:49:25.061173 131416834704960 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 01:49:25.061334 131543128213312 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 01:49:25.061489 131543128213312 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:49:25.062707 131543128213312 metric_logger.py:196] completed step: 9, seconds: 0.052, TFLOP/s/device: 64.730, Tokens/s/device: 9756.841, total_weights: 3004, loss: 8.888, lm_loss: 8.888, perplexity: 7242.487 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops