2026-04-17 13:29:17.259079: 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) I0417 13:29:17.374158 137454873017472 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:30:04.843314 137454873017472 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:30:04.843421 137454873017472 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:30:04.843457 137454873017472 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0417 13:30:04.843482 137454873017472 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:30:04.843506 137454873017472 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:30:04.843597 137454873017472 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:30:04.935956 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:30:05.320840 137454873017472 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:30:05.321209 137454873017472 checkpointing.py:677] Setting up checkpoint logger... I0417 13:30:05.321258 137454873017472 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:30:05.321313 137454873017472 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:30:05.321832 137454873017472 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 0x7d031f3ff230>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:30:07.857452 137454873017472 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:30:07.857792 137454873017472 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 0x7cfca1d32270>}, handler_registry=None I0417 13:30:07.858216 137454873017472 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cfca1d32270>` 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`. I0417 13:30:07.858262 137454873017472 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cfe1fcb8440>` 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`. I0417 13:30:07.858296 137454873017472 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 0x7cfca1d32270>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cfca1d32270>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cfe1fcb8440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cfe1fcb8440>}). I0417 13:30:07.858693 137454873017472 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:30:07.858762 137454873017472 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7cfca1d2e020> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:30:07.981376 137454873017472 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints I0417 13:30:07.981615 137454873017472 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_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7cfca1d32510> I0417 13:30:07.981715 137454873017472 checkpointing.py:301] Checkpoint manager created! W0417 13:30:07.999370 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:30:08.536318 137454873017472 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:30:08.536418 137454873017472 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:30:08.620659 137454873017472 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:30:08.620735 137454873017472 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:30:08.635245 137454873017472 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:30:08.635294 137454873017472 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:30:08.656134 137454873017472 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:30:08.656188 137454873017472 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:30:08.670520 137454873017472 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:30:08.670571 137454873017472 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:30:08.684925 137454873017472 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0417 13:30:08.684973 137454873017472 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:30:08.706632 137454873017472 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:30:08.706689 137454873017472 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:30:08.725060 137454873017472 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:30:08.725111 137454873017472 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:30:08.920140 137454873017472 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:30:08.920233 137454873017472 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 13:30:08.970952 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 I0417 13:30:09.926717 137454873017472 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:30:09.926836 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:30:09.926885 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0417 13:30:09.926930 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0417 13:30:09.926971 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:30:09.926999 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0417 13:30:09.927057 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 13:30:09.927103 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0417 13:30:09.927133 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0417 13:30:09.927159 137454873017472 maxtext_utils.py:1651] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0417 13:30:09.927195 137454873017472 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:30:09.927229 137454873017472 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:30:10.722579 137454873017472 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:30:10.722661 137454873017472 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:30:10.736327 137454873017472 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:30:10.736376 137454873017472 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:30:18.503219 137454873017472 max_utils.py:791] Total memory size: 2.4 GB, Output size: 0.5 GB, Temp size: 1.9 GB, Argument size: 0.5 GB, Host temp size: 0.0 GB. I0417 13:30:18.504170 137454873017472 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:30:18.504332 137454873017472 metric_logger.py:301] number parameters: 1.104 billion I0417 13:30:26.851788 137454873017472 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:30:27.047191 137454873017472 checkpointing.py:776] Waited 0.1953873634338379 seconds for step 0 to finish before starting checkpointing. I0417 13:30:27.047666 137454873017472 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:30:27.047819 137454873017472 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:30:27.048193 137454873017472 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0. I0417 13:30:27.151875 137454873017472 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:30:27.238586 137347330999872 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0 I0417 13:30:27.252009 137454873017472 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:30:27.252119 137454873017472 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 W0417 13:30:27.541191 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.551126 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.568338 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.579214 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.584909 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.590026 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.595521 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.601010 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:30:27.779547 137454873017472 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.527936s I0417 13:30:27.779894 137454873017472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 19.691 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 626 milliseconds) (per-host) I0417 13:30:27.779967 137454873017472 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.626841s (batch_requests_ready=0.094701s, total_serialization_initiated=0.531846s, others=0.000295s) I0417 13:30:27.780062 137454873017472 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.627693s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.627674) I0417 13:30:27.780840 137347228239424 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:30:27.780936 137454873017472 async_checkpointer.py:561] Finished blocking save. Time taken: 0.733077s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0. I0417 13:30:27.781151 137454873017472 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:30:27.781318 137347217753664 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:30:27.781386 137454873017472 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432627.047654, 'wait_for_prev_duration_secs': 4.3392181396484375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432627.04784, 'checkpointer_blocking_duration_secs': 0.7332384586334229, 'get_old_steps_start_time': 1776432627.7810962, 'get_old_steps_duration_secs': 2.09808349609375e-05, 'checkpoint_manager_blocking_start_time': 1776432627.047566, 'checkpoint_manager_blocking_duration_secs': 0.7337911128997803} I0417 13:30:27.781499 137454873017472 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:30:27.781548 137454873017472 max_utils.py:750] Memstats: After params initialized: I0417 13:30:27.781627 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:30:27.781652 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:30:27.781672 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:30:27.781691 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:30:27.781708 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:30:27.781730 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:30:27.781752 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:30:27.781772 137454873017472 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:30:27.787706 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.793874 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:30:27.799376 467419 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:30:27.966522 137347270182464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/items I0417 13:30:28.168087 137347341485632 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432627869831481, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776432627869831481, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:30:28.176441 137454873017472 metric_logger.py:196] completed step: 0, seconds: 8.346, TFLOP/s/device: 1.628, Tokens/s/device: 245.376, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.840 I0417 13:30:28.177110 137454873017472 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/tensorboard/' I0417 13:30:28.357832 137454873017472 metric_logger.py:196] completed step: 1, seconds: 1.322, TFLOP/s/device: 10.281, Tokens/s/device: 1549.627, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.840 I0417 13:30:28.544261 137454873017472 metric_logger.py:196] completed step: 2, seconds: 0.008, TFLOP/s/device: 1738.596, Tokens/s/device: 262060.141, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.299 I0417 13:30:28.579311 470519 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:30:28.579371 470519 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:30:28.729218 137454873017472 metric_logger.py:196] completed step: 3, seconds: 0.182, TFLOP/s/device: 74.459, Tokens/s/device: 11223.209, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.454 I0417 13:30:29.406824 137347312125504 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_0 I0417 13:30:49.873612 137454873017472 metric_logger.py:196] completed step: 4, seconds: 0.186, TFLOP/s/device: 72.876, Tokens/s/device: 10984.590, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2879.114 I0417 13:30:49.882381 137454873017472 metric_logger.py:196] completed step: 5, seconds: 0.185, TFLOP/s/device: 73.504, Tokens/s/device: 11079.373, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.691 I0417 13:30:50.065388 137454873017472 metric_logger.py:196] completed step: 6, seconds: 21.146, TFLOP/s/device: 0.643, Tokens/s/device: 96.851, total_weights: 16384, loss: 6.677, lm_loss: 6.677, perplexity: 793.605 I0417 13:30:50.250503 137454873017472 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2742.658, Tokens/s/device: 413403.310, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.987 I0417 13:30:50.436194 137454873017472 metric_logger.py:196] completed step: 8, seconds: 0.184, TFLOP/s/device: 73.825, Tokens/s/device: 11127.713, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.449 I0417 13:30:50.620931 137454873017472 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:30:50.622014 137454873017472 checkpointing.py:776] Waited 0.0011115074157714844 seconds for step 9 to finish before starting checkpointing. I0417 13:30:50.622304 137454873017472 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:31:01.231269 137347238725184 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.658977s (commit=32.187278s, array_metadata_write=0.471698s) I0417 13:31:01.232397 137347228239424 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 370.806 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host) I0417 13:31:01.232453 137347228239424 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.451442s. I0417 13:31:01.476481 137347228239424 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432627869831481, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:31:01.651558 137347228239424 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:31:01.857232 137347341485632 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432627869831481, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:31:02.104659 137347228239424 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:31:02.105296 137347228239424 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.590845s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/items I0417 13:31:02.105994 137347228239424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0/items I0417 13:31:02.357993 137347228239424 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0 I0417 13:31:03.065953 137347228239424 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0`. I0417 13:31:03.066590 137347228239424 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 36.018738s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/0 I0417 13:31:03.066669 137347228239424 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.285658s. I0417 13:31:03.066795 137347217753664 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:31:03.066846 137347217753664 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:31:03.066901 137347217753664 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:31:03.066942 137347217753664 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:31:03.067115 137454873017472 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. W0417 13:31:03.067248 137454873017472 checkpoint_manager.py:1441] Waiting for previous save to complete took 12.444956 seconds. If this number is high, consider checkpointing less frequently. I0417 13:31:03.068220 137454873017472 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:31:03.068509 137454873017472 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9. I0417 13:31:03.245327 137454873017472 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:31:03.245414 137454873017472 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 I0417 13:31:03.323146 137347217753664 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9 I0417 13:31:03.491040 137454873017472 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.246182s I0417 13:31:03.491456 137454873017472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 37.533 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 328 milliseconds) (per-host) I0417 13:31:03.491505 137454873017472 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.328879s (batch_requests_ready=0.079440s, total_serialization_initiated=0.249135s, others=0.000304s) I0417 13:31:03.491562 137454873017472 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.329382s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.329364) I0417 13:31:03.492341 137347182102080 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:31:03.492439 137454873017472 async_checkpointer.py:561] Finished blocking save. Time taken: 0.424172s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9. I0417 13:31:03.492616 137454873017472 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:31:03.492795 137347171616320 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:31:03.492913 137454873017472 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432650.6222615, 'wait_for_prev_duration_secs': 12.444955825805664, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432663.0682466, 'checkpointer_blocking_duration_secs': 0.42429089546203613, 'get_old_steps_start_time': 1776432663.4925525, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776432650.622216, 'checkpoint_manager_blocking_duration_secs': 12.870665788650513} I0417 13:31:03.493124 137454873017472 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:31:03.493159 137454873017472 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:31:04.028191 137347270182464 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9/items I0417 13:31:05.294234 137347259696704 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_0 I0417 13:31:41.191474 137347192587840 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.672417s (commit=36.239913s, array_metadata_write=0.432504s) I0417 13:31:41.192596 137347182102080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 332.286 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0417 13:31:41.192711 137347182102080 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.700225s. I0417 13:31:41.678315 137347182102080 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0417 13:31:42.142662 137347182102080 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:31:42.142833 137347182102080 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.618531s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9/items I0417 13:31:42.143454 137347182102080 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9/items I0417 13:31:42.388794 137347182102080 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9 I0417 13:31:43.085927 137347182102080 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9`. I0417 13:31:43.086561 137347182102080 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.018300s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_true/checkpoints/9 I0417 13:31:43.086646 137347182102080 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.594161s. I0417 13:31:43.086831 137347171616320 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:31:43.086946 137347171616320 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:31:43.087006 137347171616320 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:31:43.087061 137347171616320 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:31:43.087216 137454873017472 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. I0417 13:31:43.087352 137454873017472 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:31:43.088102 137454873017472 metric_logger.py:196] completed step: 9, seconds: 0.185, TFLOP/s/device: 73.499, Tokens/s/device: 11078.534, total_weights: 16384, loss: 5.915, lm_loss: 5.915, perplexity: 370.686 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops