2026-04-22 18:23:00.274979: 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 00:00:1776882180.288314 2058861 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776882180.292191 2058861 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776882180.303743 2058861 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776882180.303760 2058861 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776882180.303762 2058861 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776882180.303763 2058861 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 18:23:03.989754: 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 18:23:04.125056 126235024291648 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 18:23:37.377534 126235024291648 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:23:37.377839 126235024291648 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:23:37.377890 126235024291648 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 18:23:37.377919 126235024291648 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 18:23:37.377942 126235024291648 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:23:37.378126 126235024291648 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 18:23:38.038856 126235024291648 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:23:38.039158 126235024291648 checkpointing.py:677] Setting up checkpoint logger... I0422 18:23:38.039206 126235024291648 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 18:23:38.039243 126235024291648 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 18:23:38.039669 126235024291648 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 0x72cf183aad20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:23:40.583106 126235024291648 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 18:23:40.583337 126235024291648 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 0x72cdbeb5a7b0>}, handler_registry=None I0422 18:23:40.583568 126235024291648 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x72cdbeb5a7b0>` 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 18:23:40.583609 126235024291648 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72cd3f356000>` 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 18:23:40.583639 126235024291648 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 0x72cdbeb5a7b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x72cdbeb5a7b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72cd3f356000>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x72cd3f356000>}). I0422 18:23:40.583907 126235024291648 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 18:23:40.583961 126235024291648 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x72cd3f368680> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:23:40.715795 126235024291648 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints I0422 18:23:40.716040 126235024291648 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_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x72cdbc3b36e0> I0422 18:23:40.716331 126235024291648 checkpointing.py:301] Checkpoint manager created! I0422 18:23:41.095438 126235024291648 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:23:41.095539 126235024291648 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:23:41.374714 126235024291648 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:23:41.374809 126235024291648 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:23:41.387022 126235024291648 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:23:41.387084 126235024291648 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:23:41.405933 126235024291648 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 18:23:41.405999 126235024291648 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:23:41.418237 126235024291648 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 18:23:41.418291 126235024291648 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:23:41.430444 126235024291648 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 18:23:41.430496 126235024291648 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:23:41.449813 126235024291648 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:23:41.449878 126235024291648 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:23:41.465252 126235024291648 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:23:41.465308 126235024291648 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 18:23:41.614497 126235024291648 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:23:41.614600 126235024291648 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 18:23:42.229278 126235024291648 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 18:23:42.229393 126235024291648 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 18:23:42.229436 126235024291648 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 18:23:42.229482 126235024291648 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 18:23:42.229522 126235024291648 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 18:23:42.229555 126235024291648 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 18:23:42.229597 126235024291648 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 18:23:42.229638 126235024291648 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 18:23:42.229668 126235024291648 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 18:23:42.229695 126235024291648 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 18:23:42.229737 126235024291648 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 18:23:42.229773 126235024291648 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 18:23:42.636255 126235024291648 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:23:42.636350 126235024291648 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:23:42.647814 126235024291648 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:23:42.647864 126235024291648 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:23:43.404347 126235024291648 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 18:23:43.404914 126235024291648 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 18:23:43.405083 126235024291648 metric_logger.py:301] number parameters: 1.104 billion I0422 18:23:46.424597 126235024291648 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 18:23:46.542655 126235024291648 checkpointing.py:776] Waited 0.11804413795471191 seconds for step 0 to finish before starting checkpointing. I0422 18:23:46.543319 126235024291648 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:23:46.543472 126235024291648 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 18:23:46.543873 126235024291648 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0. I0422 18:23:46.625273 126235024291648 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 18:23:46.720669 126235024291648 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:23:46.720765 126235024291648 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 18:23:46.729405 126203946075712 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0 I0422 18:23:47.461727 126203935589952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/items I0422 18:23:47.615101 126203549713984 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776882227364824498, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776882227364824498, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0422 18:23:47.816117 126235024291648 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.095886s I0422 18:23:47.816528 126235024291648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 10.372 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0422 18:23:47.816586 126235024291648 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.189934s (batch_requests_ready=0.089709s, total_serialization_initiated=1.099943s, others=0.000281s) I0422 18:23:47.816653 126235024291648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.190842s (all_items=0.000022s, per_item={'items': '0.00002217'}, temp_paths=1.190820) I0422 18:23:47.817634 126203329513024 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:23:47.817765 126235024291648 async_checkpointer.py:561] Finished blocking save. Time taken: 1.274248s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0. I0422 18:23:47.818097 126235024291648 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:23:47.818491 126203956561472 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:23:47.818618 126235024291648 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776882226.5433056, 'wait_for_prev_duration_secs': 4.291534423828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776882226.5434928, 'checkpointer_blocking_duration_secs': 1.2744154930114746, 'get_old_steps_start_time': 1776882227.81793, 'get_old_steps_duration_secs': 9.393692016601562e-05, 'checkpoint_manager_blocking_start_time': 1776882226.5432243, 'checkpoint_manager_blocking_duration_secs': 1.2753593921661377} I0422 18:23:47.818814 126235024291648 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 18:23:47.818863 126235024291648 max_utils.py:750] Memstats: After params initialized: I0422 18:23:47.818983 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 18:23:47.819012 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 18:23:47.819036 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 18:23:47.819072 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 18:23:47.819092 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 18:23:47.819112 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 18:23:47.819132 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 18:23:47.819150 126235024291648 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 18:23:47.932366 2060954 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 18:23:47.932420 2060954 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 18:23:48.181362 126235024291648 metric_logger.py:196] completed step: 0, seconds: 3.019, TFLOP/s/device: 4.501, Tokens/s/device: 678.376, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 18:23:48.182438 126235024291648 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/tensorboard/' I0422 18:23:48.297864 126235024291648 metric_logger.py:196] completed step: 1, seconds: 1.754, TFLOP/s/device: 7.744, Tokens/s/device: 1167.291, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 18:23:48.417740 126235024291648 metric_logger.py:196] completed step: 2, seconds: 0.008, TFLOP/s/device: 1788.722, Tokens/s/device: 269615.587, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0422 18:23:48.536607 126235024291648 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.572, Tokens/s/device: 17420.299, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0422 18:23:48.715088 126203539228224 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_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0422 18:24:05.257515 126235024291648 metric_logger.py:196] completed step: 4, seconds: 0.120, TFLOP/s/device: 113.544, Tokens/s/device: 17114.588, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0422 18:24:05.283604 126235024291648 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.109, Tokens/s/device: 17199.822, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0422 18:24:05.402919 126235024291648 metric_logger.py:196] completed step: 6, seconds: 16.739, TFLOP/s/device: 0.812, Tokens/s/device: 122.352, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0422 18:24:05.525250 126235024291648 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1929.168, Tokens/s/device: 290785.177, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0422 18:24:05.644704 126235024291648 metric_logger.py:196] completed step: 8, seconds: 0.121, TFLOP/s/device: 112.663, Tokens/s/device: 16981.758, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0422 18:24:05.763182 126235024291648 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0422 18:24:05.765634 126235024291648 checkpointing.py:776] Waited 0.002719879150390625 seconds for step 9 to finish before starting checkpointing. I0422 18:24:05.771381 126235024291648 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:24:52.530806 126203786692160 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 64.602074s (commit=64.116842s, array_metadata_write=0.485232s) I0422 18:24:52.532033 126203329513024 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 191.741 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host) I0422 18:24:52.532111 126203329513024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 64.714282s. I0422 18:24:52.798378 126203329513024 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776882227364824498, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0422 18:24:53.026628 126203329513024 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 18:24:53.252006 126203549713984 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776882227364824498, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0422 18:24:53.457354 126203329513024 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:24:53.457575 126203329513024 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.610084s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/items I0422 18:24:53.458390 126203329513024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0/items I0422 18:24:53.699978 126203329513024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0 I0422 18:24:54.440090 126203329513024 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_11_optimizer_offload_false/checkpoints/0`. I0422 18:24:54.440846 126203329513024 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 67.897338s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/0 I0422 18:24:54.440936 126203329513024 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 66.623111s. I0422 18:24:54.441187 126203956561472 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:24:54.441329 126203956561472 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:24:54.441403 126203956561472 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:24:54.441459 126203956561472 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 18:24:54.441658 126235024291648 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 18:24:54.441911 126235024291648 checkpoint_manager.py:1441] Waiting for previous save to complete took 48.674278 seconds. If this number is high, consider checkpointing less frequently. I0422 18:24:54.443237 126235024291648 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0422 18:24:54.443516 126235024291648 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9. I0422 18:24:54.619539 126203329513024 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9 I0422 18:24:54.625225 126235024291648 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:24:54.625330 126235024291648 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 18:24:55.323003 126202945734208 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9/items I0422 18:24:55.513709 126235024291648 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.888936s I0422 18:24:55.514224 126235024291648 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.626 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 977 milliseconds) (per-host) I0422 18:24:55.514279 126235024291648 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.977534s (batch_requests_ready=0.085105s, total_serialization_initiated=0.892031s, others=0.000398s) I0422 18:24:55.514342 126235024291648 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.978111s (all_items=0.000023s, per_item={'items': '0.00002337'}, temp_paths=0.978087) I0422 18:24:55.515423 126203744749120 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:24:55.515529 126235024291648 async_checkpointer.py:561] Finished blocking save. Time taken: 1.072244s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9. I0422 18:24:55.515757 126235024291648 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:24:55.516039 126203956561472 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:24:55.516174 126235024291648 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776882245.7675304, 'wait_for_prev_duration_secs': 48.674278020858765, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776882294.4432638, 'checkpointer_blocking_duration_secs': 1.0723762512207031, 'get_old_steps_start_time': 1776882295.5156565, 'get_old_steps_duration_secs': 5.4836273193359375e-05, 'checkpoint_manager_blocking_start_time': 1776882245.765932, 'checkpoint_manager_blocking_duration_secs': 49.75021743774414} I0422 18:24:55.516290 126235024291648 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0422 18:24:55.516325 126235024291648 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:24:56.593009 126203755234880 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_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0422 18:25:30.254159 126203786692160 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 34.443671s (commit=33.785446s, array_metadata_write=0.658225s) I0422 18:25:30.255390 126203744749120 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 353.787 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 35 seconds) (per-host) I0422 18:25:30.255517 126203744749120 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 34.739937s. I0422 18:25:30.787679 126203744749120 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 18:25:31.180162 126203744749120 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:25:31.180349 126203744749120 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.594150s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9/items I0422 18:25:31.181244 126203744749120 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9/items I0422 18:25:31.413748 126203744749120 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9 I0422 18:25:32.117721 126203744749120 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_11_optimizer_offload_false/checkpoints/9`. I0422 18:25:32.118478 126203744749120 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 37.675198s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_11_optimizer_offload_false/checkpoints/9 I0422 18:25:32.118582 126203744749120 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 36.603004s. I0422 18:25:32.118827 126203956561472 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:25:32.118955 126203956561472 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:25:32.119007 126203956561472 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:25:32.119067 126203956561472 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 18:25:32.119259 126235024291648 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 18:25:32.119411 126235024291648 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:25:32.120427 126235024291648 metric_logger.py:196] completed step: 9, seconds: 0.124, TFLOP/s/device: 109.624, Tokens/s/device: 16523.725, total_weights: 16384, loss: 5.915, lm_loss: 5.915, perplexity: 370.653 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops