2026-04-22 19:04:08.091270: 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 00:00:1776884648.104318 2097094 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776884648.108149 2097094 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776884648.119617 2097094 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884648.119631 2097094 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884648.119632 2097094 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776884648.119635 2097094 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 19:04:12.023844: 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 19:04:12.161754 137645457684288 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 19:04:43.400007 137645457684288 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 19:04:43.400138 137645457684288 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 19:04:43.400172 137645457684288 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 19:04:43.400203 137645457684288 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 19:04:43.400227 137645457684288 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 19:04:43.400318 137645457684288 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 19:04:44.111483 137645457684288 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 19:04:44.111779 137645457684288 checkpointing.py:677] Setting up checkpoint logger... I0422 19:04:44.111824 137645457684288 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0422 19:04:44.111861 137645457684288 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 19:04:44.112350 137645457684288 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7d2fcd9b1a60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 19:04:46.614853 137645457684288 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 19:04:46.615268 137645457684288 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 0x7d2df3f4d6a0>}, handler_registry=None I0422 19:04:46.615516 137645457684288 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2df3f4d6a0>` 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 19:04:46.615558 137645457684288 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2df3f4dc10>` 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 19:04:46.615588 137645457684288 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 0x7d2df3f4d6a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d2df3f4d6a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2df3f4dc10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d2df3f4dc10>}). I0422 19:04:46.615878 137645457684288 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 19:04:46.615936 137645457684288 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7d2df3f5c900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 19:04:46.748784 137645457684288 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints I0422 19:04:46.749028 137645457684288 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=5), 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_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d2df3f4cb30> I0422 19:04:46.749332 137645457684288 checkpointing.py:301] Checkpoint manager created! I0422 19:04:47.133584 137645457684288 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 19:04:47.133684 137645457684288 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:04:47.416341 137645457684288 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 19:04:47.416438 137645457684288 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:04:47.428781 137645457684288 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 19:04:47.428833 137645457684288 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 19:04:47.447997 137645457684288 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 19:04:47.448074 137645457684288 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:04:47.460509 137645457684288 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 19:04:47.460562 137645457684288 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:04:47.472883 137645457684288 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 19:04:47.472932 137645457684288 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:04:47.492358 137645457684288 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 19:04:47.492419 137645457684288 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 19:04:47.507894 137645457684288 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 19:04:47.507953 137645457684288 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 19:04:47.659109 137645457684288 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 19:04:47.659211 137645457684288 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 19:04:48.286445 137645457684288 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 19:04:48.286557 137645457684288 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 19:04:48.286599 137645457684288 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 19:04:48.286643 137645457684288 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 19:04:48.286688 137645457684288 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 19:04:48.286717 137645457684288 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 19:04:48.286758 137645457684288 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 19:04:48.286807 137645457684288 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 19:04:48.286836 137645457684288 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 19:04:48.286862 137645457684288 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 19:04:48.286898 137645457684288 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 19:04:48.286934 137645457684288 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 19:04:48.695559 137645457684288 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 19:04:48.695653 137645457684288 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 19:04:48.707294 137645457684288 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 19:04:48.707347 137645457684288 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 19:04:49.469005 137645457684288 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. I0422 19:04:49.469682 137645457684288 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 19:04:49.469852 137645457684288 metric_logger.py:301] number parameters: 1.104 billion I0422 19:04:51.798232 137645457684288 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 19:04:51.915779 137645457684288 checkpointing.py:776] Waited 0.11754798889160156 seconds for step 0 to finish before starting checkpointing. I0422 19:04:51.916580 137645457684288 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 19:04:51.916772 137645457684288 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 19:04:51.917327 137645457684288 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0. I0422 19:04:52.006654 137645457684288 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 19:04:52.097394 137645457684288 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:04:52.097494 137645457684288 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 19:04:52.104113 137614363461184 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0 I0422 19:04:52.864674 137614352975424 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/items I0422 19:04:53.033698 137614321518144 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884692767781160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776884692767781160, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0422 19:04:53.359626 2099189 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 19:04:53.359690 2099189 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 19:04:54.497938 137645457684288 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.401030s I0422 19:04:54.498681 137645457684288 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.954 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 19:04:54.498739 137645457684288 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.490971s (batch_requests_ready=0.085114s, total_serialization_initiated=2.405242s, others=0.000615s) I0422 19:04:54.498807 137645457684288 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.491601s (all_items=0.000024s, per_item={'items': '0.00002408'}, temp_paths=2.491576) I0422 19:04:54.504033 137613723829824 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:04:54.504149 137645457684288 async_checkpointer.py:561] Finished blocking save. Time taken: 2.587333s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0. I0422 19:04:54.504356 137645457684288 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:04:54.504512 137614373946944 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:04:54.504576 137645457684288 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884691.9165645, 'wait_for_prev_duration_secs': 4.4345855712890625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884691.9167943, 'checkpointer_blocking_duration_secs': 2.587480068206787, 'get_old_steps_start_time': 1776884694.504291, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776884691.9164422, 'checkpoint_manager_blocking_duration_secs': 2.588108539581299} I0422 19:04:54.504688 137645457684288 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 19:04:54.504733 137645457684288 max_utils.py:750] Memstats: After params initialized: I0422 19:04:54.504783 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 19:04:54.504807 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 19:04:54.504826 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 19:04:54.504845 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 19:04:54.504862 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 19:04:54.504878 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 19:04:54.504894 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 19:04:54.504910 137645457684288 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 19:04:54.756872 137645457684288 metric_logger.py:196] completed step: 0, seconds: 2.328, TFLOP/s/device: 5.837, Tokens/s/device: 879.839, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 19:04:54.757775 137645457684288 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/tensorboard/' I0422 19:04:54.873935 137645457684288 metric_logger.py:196] completed step: 1, seconds: 2.957, TFLOP/s/device: 4.595, Tokens/s/device: 692.664, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 19:04:54.993185 137645457684288 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2150.543, Tokens/s/device: 324153.213, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0422 19:04:55.111531 137645457684288 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.432, Tokens/s/device: 17399.135, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0422 19:04:55.116981 137645457684288 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0422 19:04:55.296145 137614332003904 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_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0422 19:04:55.348650 137645457684288 checkpointing.py:776] Waited 0.23163199424743652 seconds for step 5 to finish before starting checkpointing. I0422 19:04:55.349587 137645457684288 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:05:32.346814 137613734315584 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.843544s (commit=37.370568s, array_metadata_write=0.472977s) I0422 19:05:32.348011 137613723829824 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 313.256 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0422 19:05:32.348145 137613723829824 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.843946s. I0422 19:05:32.599481 137613723829824 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884692767781160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0422 19:05:32.841723 137613723829824 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 19:05:33.047447 137614321518144 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776884692767781160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0422 19:05:33.358096 137613723829824 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.707060s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/items I0422 19:05:33.358926 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0/items I0422 19:05:33.620746 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0 I0422 19:05:34.354200 137613723829824 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_16_zarr3_false/checkpoints/0`. I0422 19:05:34.354938 137613723829824 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.438128s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/0 I0422 19:05:34.355035 137613723829824 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.850839s. I0422 19:05:34.355174 137614373946944 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:05:34.355225 137614373946944 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:05:34.355285 137614373946944 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:05:34.355326 137614373946944 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 19:05:34.355498 137645457684288 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 19:05:34.355637 137645457684288 checkpoint_manager.py:1441] Waiting for previous save to complete took 39.006053 seconds. If this number is high, consider checkpointing less frequently. I0422 19:05:34.356410 137645457684288 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0422 19:05:34.356725 137645457684288 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5. I0422 19:05:34.523953 137613723829824 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5 I0422 19:05:34.528551 137645457684288 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:05:34.528653 137645457684288 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 19:05:35.275680 137614332003904 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5/items I0422 19:05:36.580995 137645457684288 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.052898s I0422 19:05:36.585029 137645457684288 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.754 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 19:05:36.585105 137645457684288 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.144832s (batch_requests_ready=0.084801s, total_serialization_initiated=2.056135s, others=0.003896s) I0422 19:05:36.585208 137645457684288 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.145409s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=2.145394) I0422 19:05:36.586719 137613723829824 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:05:36.586808 137645457684288 async_checkpointer.py:561] Finished blocking save. Time taken: 2.230354s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5. I0422 19:05:36.586999 137645457684288 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:05:36.587183 137614373946944 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:05:36.587303 137645457684288 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884695.3495631, 'wait_for_prev_duration_secs': 39.00605273246765, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776884734.356434, 'checkpointer_blocking_duration_secs': 2.230485677719116, 'get_old_steps_start_time': 1776884736.586934, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1776884695.3494973, 'checkpoint_manager_blocking_duration_secs': 41.23778033256531} I0422 19:05:36.587455 137645457684288 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0422 19:05:36.588246 137645457684288 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.741, Tokens/s/device: 17144.244, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0422 19:05:36.596341 137645457684288 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 114.755, Tokens/s/device: 17297.151, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0422 19:05:36.712818 137645457684288 metric_logger.py:196] completed step: 6, seconds: 41.477, TFLOP/s/device: 0.328, Tokens/s/device: 49.377, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0422 19:05:36.831760 137645457684288 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2006.961, Tokens/s/device: 302511.078, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0422 19:05:36.950623 137645457684288 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.890, Tokens/s/device: 17468.143, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0422 19:05:37.068896 137645457684288 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0422 19:05:37.069797 137645457684288 checkpointing.py:776] Waited 0.0009264945983886719 seconds for step 9 to finish before starting checkpointing. I0422 19:05:37.070088 137645457684288 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:05:37.328901 137614172620352 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_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0422 19:06:08.617955 137614332003904 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.032097s (commit=31.553828s, array_metadata_write=0.478269s) I0422 19:06:08.619078 137613723829824 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 369.726 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0422 19:06:08.619135 137613723829824 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.032283s. I0422 19:06:09.158685 137613723829824 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 19:06:09.633755 137613723829824 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.652152s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5/items I0422 19:06:09.634591 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5/items I0422 19:06:09.875218 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5 I0422 19:06:10.609545 137613723829824 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_16_zarr3_false/checkpoints/5`. I0422 19:06:10.610275 137613723829824 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.253827s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/5 I0422 19:06:10.610349 137613723829824 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.023498s. I0422 19:06:10.610534 137614373946944 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:06:10.610656 137614373946944 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:06:10.610727 137614373946944 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:06:10.610771 137614373946944 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0422 19:06:10.610905 137645457684288 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0422 19:06:10.611027 137645457684288 checkpoint_manager.py:1441] Waiting for previous save to complete took 33.540950 seconds. If this number is high, consider checkpointing less frequently. I0422 19:06:10.611340 137645457684288 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0422 19:06:10.611635 137645457684288 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9. I0422 19:06:10.787230 137613723829824 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9 I0422 19:06:10.787400 137645457684288 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 19:06:10.787632 137645457684288 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 19:06:11.511601 137614172620352 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9/items I0422 19:06:16.430148 137645457684288 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.643522s I0422 19:06:16.430742 137645457684288 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.153 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0422 19:06:16.430793 137645457684288 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.731049s (batch_requests_ready=0.083831s, total_serialization_initiated=5.646777s, others=0.000441s) I0422 19:06:16.430855 137645457684288 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.731538s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=5.731526) I0422 19:06:16.435546 137613723829824 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 19:06:16.435716 137645457684288 async_checkpointer.py:561] Finished blocking save. Time taken: 5.824335s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9. I0422 19:06:16.435843 137645457684288 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 19:06:16.436096 137614126483008 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 19:06:16.436216 137645457684288 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776884737.0700626, 'wait_for_prev_duration_secs': 33.54094982147217, 'time_between_consecutive_saves_sec': 2.714719533920288, 'checkpointer_blocking_start_time': 1776884770.6113636, 'checkpointer_blocking_duration_secs': 5.824412107467651, 'get_old_steps_start_time': 1776884776.4357927, 'get_old_steps_duration_secs': 1.71661376953125e-05, 'checkpoint_manager_blocking_start_time': 1776884737.070017, 'checkpoint_manager_blocking_duration_secs': 39.36617350578308} I0422 19:06:16.436330 137645457684288 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0422 19:06:16.436365 137645457684288 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 19:06:17.218766 137614162134592 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_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0422 19:06:55.903629 137614151648832 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 39.468646s (commit=38.996362s, array_metadata_write=0.472283s) I0422 19:06:55.904755 137613723829824 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 279.544 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host) I0422 19:06:55.904881 137613723829824 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.469246s. I0422 19:06:56.402624 137613723829824 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 19:06:56.958653 137613723829824 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.728305s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9/items I0422 19:06:56.959494 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9/items I0422 19:06:57.213317 137613723829824 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9 I0422 19:06:57.922567 137613723829824 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_16_zarr3_false/checkpoints/9`. I0422 19:06:57.923310 137613723829824 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.311935s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_16_zarr3_false/checkpoints/9 I0422 19:06:57.923376 137613723829824 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.487742s. I0422 19:06:57.923532 137614126483008 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 19:06:57.923653 137614126483008 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 19:06:57.923703 137614126483008 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 19:06:57.923745 137614126483008 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 19:06:57.923914 137645457684288 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 19:06:57.924070 137645457684288 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 19:06:57.925016 137645457684288 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.056, Tokens/s/device: 17191.737, total_weights: 16384, loss: 4.189, lm_loss: 4.189, perplexity: 65.979 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops