2026-04-22 18:45:19.765237: 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 00:00:1776883519.778502 2078901 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776883519.782343 2078901 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776883519.793806 2078901 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883519.793822 2078901 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883519.793824 2078901 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776883519.793825 2078901 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 18:45:23.463448: 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 18:45:23.598960 136422333294400 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 18:45:53.782915 136422333294400 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:45:53.783197 136422333294400 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:45:53.783253 136422333294400 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 18:45:53.783283 136422333294400 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 18:45:53.783308 136422333294400 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:45:53.783475 136422333294400 train.py:705] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 18:45:54.441748 136422333294400 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:45:54.442021 136422333294400 checkpointing.py:677] Setting up checkpoint logger... I0422 18:45:54.442090 136422333294400 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 18:45:54.442127 136422333294400 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 18:45:54.442555 136422333294400 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 0x7c1305badcd0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:45:56.887540 136422333294400 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 18:45:56.887910 136422333294400 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 0x7c112f34db20>}, handler_registry=None I0422 18:45:56.888169 136422333294400 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c112f34db20>` 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 18:45:56.888211 136422333294400 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c112f34d010>` 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 18:45:56.888242 136422333294400 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 0x7c112f34db20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c112f34db20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c112f34d010>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c112f34d010>}). I0422 18:45:56.888511 136422333294400 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 18:45:56.888563 136422333294400 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7c112f370900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:45:57.010877 136422333294400 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints I0422 18:45:57.011146 136422333294400 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_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c112f34d7f0> I0422 18:45:57.011425 136422333294400 checkpointing.py:301] Checkpoint manager created! I0422 18:45:57.393464 136422333294400 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:45:57.393567 136422333294400 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:45:57.676828 136422333294400 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:45:57.676916 136422333294400 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:45:57.689209 136422333294400 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:45:57.689259 136422333294400 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:45:57.708243 136422333294400 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 18:45:57.708301 136422333294400 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:45:57.720582 136422333294400 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 18:45:57.720637 136422333294400 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:45:57.732891 136422333294400 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 18:45:57.732943 136422333294400 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:45:57.752707 136422333294400 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:45:57.752774 136422333294400 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:45:57.768220 136422333294400 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:45:57.768276 136422333294400 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 18:45:57.918951 136422333294400 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:45:57.919071 136422333294400 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 18:45:58.540259 136422333294400 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 18:45:58.540374 136422333294400 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 18:45:58.540416 136422333294400 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 18:45:58.540461 136422333294400 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 18:45:58.540504 136422333294400 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 18:45:58.540532 136422333294400 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 18:45:58.540571 136422333294400 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 18:45:58.540611 136422333294400 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 18:45:58.540639 136422333294400 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 18:45:58.540665 136422333294400 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 18:45:58.540700 136422333294400 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 18:45:58.540735 136422333294400 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 18:45:58.948395 136422333294400 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:45:58.948487 136422333294400 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:45:58.960100 136422333294400 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:45:58.960150 136422333294400 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:45:59.717415 136422333294400 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 18:45:59.717980 136422333294400 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 18:45:59.718140 136422333294400 metric_logger.py:301] number parameters: 1.104 billion I0422 18:46:01.767758 136422333294400 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 18:46:01.885808 136422333294400 checkpointing.py:776] Waited 0.11803054809570312 seconds for step 0 to finish before starting checkpointing. I0422 18:46:01.886450 136422333294400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:46:01.886601 136422333294400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 18:46:01.887011 136422333294400 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0. I0422 18:46:01.969978 136422333294400 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 18:46:02.066329 136422333294400 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:46:02.066425 136422333294400 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 18:46:02.072340 136391262471744 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0 I0422 18:46:02.856380 136391113573952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/items I0422 18:46:03.027276 136390876595776 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883562760657456, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776883562760657456, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0422 18:46:03.355050 2080973 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 18:46:03.355110 2080973 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 18:46:04.077603 136422333294400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.011733s I0422 18:46:04.082712 136422333294400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.845 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 18:46:04.082791 136422333294400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.111408s (batch_requests_ready=0.090564s, total_serialization_initiated=2.015849s, others=0.004994s) I0422 18:46:04.082878 136422333294400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.112348s (all_items=0.000022s, per_item={'items': '0.00002193'}, temp_paths=2.112326) I0422 18:46:04.084390 136390071289408 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:46:04.084476 136422333294400 async_checkpointer.py:561] Finished blocking save. Time taken: 2.197834s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0. I0422 18:46:04.084669 136422333294400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:46:04.084883 136391272957504 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:46:04.085003 136422333294400 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883561.8864374, 'wait_for_prev_duration_secs': 4.267692565917969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883561.886623, 'checkpointer_blocking_duration_secs': 2.197964668273926, 'get_old_steps_start_time': 1776883564.0846057, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776883561.8863528, 'checkpoint_manager_blocking_duration_secs': 2.1986217498779297} I0422 18:46:04.085134 136422333294400 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 18:46:04.085178 136422333294400 max_utils.py:750] Memstats: After params initialized: I0422 18:46:04.085226 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0422 18:46:04.085251 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 18:46:04.085271 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 18:46:04.085289 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 18:46:04.085307 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 18:46:04.085324 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 18:46:04.085341 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 18:46:04.085357 136422333294400 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 18:46:04.330610 136422333294400 metric_logger.py:196] completed step: 0, seconds: 2.049, TFLOP/s/device: 6.631, Tokens/s/device: 999.470, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 18:46:04.331420 136422333294400 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/tensorboard/' I0422 18:46:04.447313 136422333294400 metric_logger.py:196] completed step: 1, seconds: 2.561, TFLOP/s/device: 5.306, Tokens/s/device: 799.715, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0422 18:46:04.566270 136422333294400 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2259.626, Tokens/s/device: 340595.377, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0422 18:46:04.685498 136422333294400 metric_logger.py:196] completed step: 3, seconds: 0.118, TFLOP/s/device: 115.461, Tokens/s/device: 17403.571, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0422 18:46:04.690922 136422333294400 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0422 18:46:04.884145 136391092602432 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_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_0 I0422 18:46:04.924036 136422333294400 checkpointing.py:776] Waited 0.23306989669799805 seconds for step 5 to finish before starting checkpointing. I0422 18:46:04.925081 136422333294400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:46:39.670390 136390272616000 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.586690s (commit=35.062557s, array_metadata_write=0.524132s) I0422 18:46:39.671563 136390071289408 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 335.192 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0422 18:46:39.671621 136390071289408 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.587101s. I0422 18:46:39.924882 136390071289408 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883562760657456, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0422 18:46:40.153162 136390071289408 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 18:46:40.360349 136390876595776 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776883562760657456, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0422 18:46:40.557766 136390071289408 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:46:40.557992 136390071289408 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.590793s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/items I0422 18:46:40.558799 136390071289408 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0/items I0422 18:46:40.800008 136390071289408 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0 I0422 18:46:41.586333 136390071289408 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_15_ocdbt_true/checkpoints/0`. I0422 18:46:41.587097 136390071289408 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.700463s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/0 I0422 18:46:41.587170 136390071289408 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.502650s. I0422 18:46:41.587362 136391272957504 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:46:41.587485 136391272957504 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:46:41.587555 136391272957504 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:46:41.587602 136391272957504 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 18:46:41.587772 136422333294400 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 18:46:41.587914 136422333294400 checkpoint_manager.py:1441] Waiting for previous save to complete took 36.662833 seconds. If this number is high, consider checkpointing less frequently. I0422 18:46:41.589319 136422333294400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0422 18:46:41.589620 136422333294400 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5. I0422 18:46:41.792112 136422333294400 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:46:41.792213 136422333294400 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 18:46:41.793407 136390071289408 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5 I0422 18:46:42.556387 136391092602432 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5/items I0422 18:46:45.445657 136422333294400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.654200s I0422 18:46:45.449880 136422333294400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.286 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 18:46:45.449954 136422333294400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.756071s (batch_requests_ready=0.094031s, total_serialization_initiated=3.657938s, others=0.004102s) I0422 18:46:45.450058 136422333294400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.756635s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=3.756617) I0422 18:46:45.451733 136390272616000 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:46:45.451852 136422333294400 async_checkpointer.py:561] Finished blocking save. Time taken: 3.862490s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5. I0422 18:46:45.452000 136422333294400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:46:45.452207 136391272957504 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:46:45.452328 136422333294400 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883564.9250548, 'wait_for_prev_duration_secs': 36.662832736968994, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776883601.589343, 'checkpointer_blocking_duration_secs': 3.862565755844116, 'get_old_steps_start_time': 1776883605.451929, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776883564.9249723, 'checkpoint_manager_blocking_duration_secs': 40.52732968330383} I0422 18:46:45.452482 136422333294400 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0422 18:46:45.453182 136422333294400 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 113.965, Tokens/s/device: 17178.038, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0422 18:46:45.461801 136422333294400 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.061, Tokens/s/device: 17192.458, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0422 18:46:45.578670 136422333294400 metric_logger.py:196] completed step: 6, seconds: 40.769, TFLOP/s/device: 0.333, Tokens/s/device: 50.234, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0422 18:46:45.697518 136422333294400 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 2151.564, Tokens/s/device: 324307.205, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0422 18:46:45.816217 136422333294400 metric_logger.py:196] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.016, Tokens/s/device: 17336.539, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0422 18:46:45.934069 136422333294400 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0422 18:46:45.935026 136422333294400 checkpointing.py:776] Waited 0.0010006427764892578 seconds for step 9 to finish before starting checkpointing. I0422 18:46:45.935300 136422333294400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:46:46.211265 136391082116672 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_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_0 I0422 18:47:46.243352 136391071630912 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 60.792649s (commit=60.314249s, array_metadata_write=0.478399s) I0422 18:47:46.244634 136390272616000 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 195.765 MiB/s (total gbytes: 12.3 GiB) (time elapsed: a minute) (per-host) I0422 18:47:46.244748 136390272616000 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 60.792929s. I0422 18:47:46.780514 136390272616000 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 18:47:47.216525 136390272616000 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:47:47.216712 136390272616000 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.637284s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5/items I0422 18:47:47.217447 136390272616000 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5/items I0422 18:47:47.454303 136390272616000 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5 I0422 18:47:48.165420 136390272616000 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_15_ocdbt_true/checkpoints/5`. I0422 18:47:48.169851 136390272616000 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 66.580497s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/5 I0422 18:47:48.169932 136390272616000 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 62.718116s. I0422 18:47:48.170168 136391272957504 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:47:48.170315 136391272957504 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:47:48.170387 136391272957504 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:47:48.170431 136391272957504 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0422 18:47:48.170613 136422333294400 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 18:47:48.170751 136422333294400 checkpoint_manager.py:1441] Waiting for previous save to complete took 62.235460 seconds. If this number is high, consider checkpointing less frequently. I0422 18:47:48.171179 136422333294400 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0422 18:47:48.171484 136422333294400 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9. I0422 18:47:48.342512 136390272616000 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9 I0422 18:47:48.348139 136422333294400 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:47:48.348236 136422333294400 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 18:47:49.062803 136391071630912 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9/items I0422 18:47:50.147008 136422333294400 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.799355s I0422 18:47:50.151222 136422333294400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 6.511 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0422 18:47:50.151290 136422333294400 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.895542s (batch_requests_ready=0.088249s, total_serialization_initiated=1.803210s, others=0.004082s) I0422 18:47:50.151361 136422333294400 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.896119s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=1.896103) I0422 18:47:50.153194 136391082116672 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:47:50.153277 136422333294400 async_checkpointer.py:561] Finished blocking save. Time taken: 1.982055s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9. I0422 18:47:50.153448 136422333294400 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:47:50.153600 136391272957504 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:47:50.153700 136422333294400 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776883605.9352746, 'wait_for_prev_duration_secs': 62.23546004295349, 'time_between_consecutive_saves_sec': 4.347657918930054, 'checkpointer_blocking_start_time': 1776883668.1712053, 'checkpointer_blocking_duration_secs': 1.9821746349334717, 'get_old_steps_start_time': 1776883670.1533918, 'get_old_steps_duration_secs': 2.3603439331054688e-05, 'checkpoint_manager_blocking_start_time': 1776883605.9352438, 'checkpoint_manager_blocking_duration_secs': 64.2184329032898} I0422 18:47:50.153833 136422333294400 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0422 18:47:50.153865 136422333294400 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:47:50.922917 136391092602432 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_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_0 I0422 18:48:32.271612 136391061145152 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.119613s (commit=41.632955s, array_metadata_write=0.486659s) I0422 18:48:32.272953 136391082116672 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 287.088 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0422 18:48:32.273095 136391082116672 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.119776s. I0422 18:48:32.768289 136391082116672 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 18:48:33.166913 136391082116672 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:48:33.167121 136391082116672 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.587826s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9/items I0422 18:48:33.167895 136391082116672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9/items I0422 18:48:33.418692 136391082116672 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9 I0422 18:48:34.141771 136391082116672 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_15_ocdbt_true/checkpoints/9`. I0422 18:48:34.142542 136391082116672 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.971325s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_15_ocdbt_true/checkpoints/9 I0422 18:48:34.142608 136391082116672 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.989290s. I0422 18:48:34.142817 136391272957504 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:48:34.142930 136391272957504 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:48:34.142998 136391272957504 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:48:34.143040 136391272957504 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 18:48:34.143229 136422333294400 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 18:48:34.143386 136422333294400 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:48:34.144640 136422333294400 metric_logger.py:196] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.526, Tokens/s/device: 17262.597, 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