2026-04-17 14:05:19.345489: 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 14:05:19.462489 140551586671744 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 14:06:16.929139 140551586671744 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 14:06:16.929242 140551586671744 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 14:06:16.929275 140551586671744 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 14:06:16.929301 140551586671744 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 14:06:16.929323 140551586671744 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 14:06:16.929393 140551586671744 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 14:06:17.018527 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:17.401807 140551586671744 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 14:06:17.402074 140551586671744 checkpointing.py:677] Setting up checkpoint logger... I0417 14:06:17.402119 140551586671744 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0417 14:06:17.402159 140551586671744 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 14:06:17.402743 140551586671744 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7fd4211db860>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 14:06:19.782872 140551586671744 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 14:06:19.783208 140551586671744 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 0x7fcda2908230>}, handler_registry=None I0417 14:06:19.783473 140551586671744 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fcda2908230>` 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 14:06:19.783514 140551586671744 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fcda2535040>` 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 14:06:19.783542 140551586671744 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 0x7fcda2908230>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fcda2908230>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fcda2535040>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fcda2535040>}). I0417 14:06:19.783865 140551586671744 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 14:06:19.783914 140551586671744 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7fcda2532200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 14:06:19.902763 140551586671744 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints I0417 14:06:19.902985 140551586671744 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_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fcda290b0e0> I0417 14:06:19.903096 140551586671744 checkpointing.py:301] Checkpoint manager created! W0417 14:06:19.919828 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:20.418951 140551586671744 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 14:06:20.419153 140551586671744 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:06:20.501746 140551586671744 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 14:06:20.501820 140551586671744 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:06:20.515968 140551586671744 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 14:06:20.516015 140551586671744 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 14:06:20.536289 140551586671744 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 14:06:20.536344 140551586671744 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:06:20.550604 140551586671744 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 14:06:20.550657 140551586671744 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:06:20.564873 140551586671744 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 14:06:20.564921 140551586671744 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:06:20.586278 140551586671744 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 14:06:20.586334 140551586671744 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 14:06:20.604484 140551586671744 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 14:06:20.604534 140551586671744 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 14:06:20.796027 140551586671744 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 14:06:20.796129 140551586671744 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 14:06:20.845500 517466 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 14:06:21.225832 140551586671744 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 14:06:21.225940 140551586671744 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 14:06:21.225981 140551586671744 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 14:06:21.226025 140551586671744 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 14:06:21.226079 140551586671744 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 14:06:21.226114 140551586671744 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 14:06:21.226155 140551586671744 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 14:06:21.226195 140551586671744 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 14:06:21.226222 140551586671744 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 14:06:21.226248 140551586671744 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 14:06:21.226281 140551586671744 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 14:06:21.226316 140551586671744 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 14:06:21.645684 140551586671744 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 14:06:21.645761 140551586671744 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 14:06:21.659349 140551586671744 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 14:06:21.659396 140551586671744 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 14:06:21.915153 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:22.031252 140551586671744 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. I0417 14:06:22.031733 140551586671744 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 14:06:22.031880 140551586671744 metric_logger.py:301] number parameters: 1.104 billion W0417 14:06:22.863839 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:22.959820 140551586671744 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 14:06:23.076693 140551586671744 checkpointing.py:776] Waited 0.11686372756958008 seconds for step 0 to finish before starting checkpointing. I0417 14:06:23.077158 140551586671744 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 14:06:23.077311 140551586671744 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 14:06:23.077691 140551586671744 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0. I0417 14:06:23.160125 140551586671744 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 14:06:23.253107 140551586671744 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:06:23.253193 140551586671744 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 14:06:23.259295 140448408536640 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0 W0417 14:06:23.467854 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.476520 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.485783 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.500017 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.510876 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.520095 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.524788 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.529252 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:23.717944 140551586671744 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.465292s I0417 14:06:23.720173 140551586671744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 22.086 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 558 milliseconds) (per-host) I0417 14:06:23.720291 140551586671744 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.558965s (batch_requests_ready=0.088032s, total_serialization_initiated=0.470356s, others=0.000576s) I0417 14:06:23.720358 140551586671744 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.559796s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.559777) I0417 14:06:23.721444 140448337233472 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:06:23.721539 140551586671744 async_checkpointer.py:561] Finished blocking save. Time taken: 0.644189s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0. I0417 14:06:23.721726 140551586671744 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:06:23.721868 140448419022400 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:06:23.721935 140551586671744 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434783.0771449, 'wait_for_prev_duration_secs': 4.315376281738281e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434783.0773308, 'checkpointer_blocking_duration_secs': 0.6443121433258057, 'get_old_steps_start_time': 1776434783.7216637, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776434783.0770311, 'checkpoint_manager_blocking_duration_secs': 0.6448721885681152} I0417 14:06:23.722056 140551586671744 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 14:06:23.722096 140551586671744 max_utils.py:750] Memstats: After params initialized: I0417 14:06:23.722193 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 14:06:23.722222 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 14:06:23.722244 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 14:06:23.722264 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 14:06:23.722283 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 14:06:23.722301 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 14:06:23.722319 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 14:06:23.722337 140551586671744 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 14:06:23.728260 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.734384 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 14:06:23.739387 517466 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 14:06:23.902007 140448398050880 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/items I0417 14:06:23.937333 140551586671744 metric_logger.py:196] completed step: 0, seconds: 0.927, TFLOP/s/device: 14.650, Tokens/s/device: 2208.177, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 14:06:23.938229 140551586671744 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/tensorboard/' I0417 14:06:24.046428 140448318359104 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434783813243820, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776434783813243820, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 14:06:24.053757 140551586671744 metric_logger.py:196] completed step: 1, seconds: 0.976, TFLOP/s/device: 13.926, Tokens/s/device: 2099.079, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 14:06:24.172359 140551586671744 metric_logger.py:196] completed step: 2, seconds: 0.006, TFLOP/s/device: 2398.858, Tokens/s/device: 361581.921, total_weights: 16384, loss: 10.276, lm_loss: 10.276, perplexity: 29019.973 I0417 14:06:24.290978 140551586671744 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.992, Tokens/s/device: 17483.503, total_weights: 16384, loss: 9.282, lm_loss: 9.282, perplexity: 10739.224 I0417 14:06:24.295430 140551586671744 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0417 14:06:24.382441 519726 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 14:06:24.382499 519726 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 14:06:24.527156 140551586671744 checkpointing.py:776] Waited 0.23169398307800293 seconds for step 5 to finish before starting checkpointing. I0417 14:06:24.527537 140551586671744 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:06:25.145829 140448358204992 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_16_zarr3_false/checkpoints/0/items/array_metadatas/process_0 I0417 14:06:59.723149 140448347719232 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.340275s (commit=34.863230s, array_metadata_write=0.477044s) I0417 14:06:59.724309 140448337233472 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 345.618 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0417 14:06:59.724412 140448337233472 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.002829s. I0417 14:06:59.942407 140448337233472 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434783813243820, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 14:07:00.119195 140448337233472 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 14:07:00.320625 140448318359104 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776434783813243820, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0417 14:07:00.619659 140448337233472 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.644583s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/items I0417 14:07:00.620365 140448337233472 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0/items I0417 14:07:00.854797 140448337233472 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0 I0417 14:07:01.564543 140448337233472 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_16_zarr3_false/checkpoints/0`. I0417 14:07:01.565230 140448337233472 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.487887s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/0 I0417 14:07:01.565305 140448337233472 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.843724s. I0417 14:07:01.565422 140448419022400 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:07:01.565476 140448419022400 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:07:01.565536 140448419022400 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:07:01.565579 140448419022400 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 14:07:01.565724 140551586671744 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 14:07:01.565851 140551586671744 checkpoint_manager.py:1441] Waiting for previous save to complete took 37.038311 seconds. If this number is high, consider checkpointing less frequently. I0417 14:07:01.566700 140551586671744 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0417 14:07:01.566987 140551586671744 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5. I0417 14:07:01.743170 140448419022400 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5 I0417 14:07:01.755702 140551586671744 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:07:01.755792 140551586671744 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 14:07:02.465430 140448276416064 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5/items I0417 14:07:10.218645 140551586671744 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.463360s I0417 14:07:10.218980 140551586671744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.442 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0417 14:07:10.219095 140551586671744 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.557752s (batch_requests_ready=0.090591s, total_serialization_initiated=8.466870s, others=0.000290s) I0417 14:07:10.219160 140551586671744 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.558267s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=8.558249) I0417 14:07:10.223329 140448419022400 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:07:10.223431 140551586671744 async_checkpointer.py:561] Finished blocking save. Time taken: 8.656686s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5. I0417 14:07:10.223610 140551586671744 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:07:10.223783 140448337233472 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:07:10.223886 140551586671744 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434784.5275152, 'wait_for_prev_duration_secs': 37.03831076622009, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776434821.5667233, 'checkpointer_blocking_duration_secs': 8.656807661056519, 'get_old_steps_start_time': 1776434830.2235484, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776434784.5274527, 'checkpoint_manager_blocking_duration_secs': 45.69641137123108} I0417 14:07:10.224010 140551586671744 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0417 14:07:10.224594 140551586671744 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.543, Tokens/s/device: 17265.217, total_weights: 16384, loss: 8.327, lm_loss: 8.327, perplexity: 4134.812 I0417 14:07:10.231656 140551586671744 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 114.765, Tokens/s/device: 17298.612, total_weights: 16384, loss: 7.367, lm_loss: 7.367, perplexity: 1583.522 I0417 14:07:10.347817 140551586671744 metric_logger.py:196] completed step: 6, seconds: 45.935, TFLOP/s/device: 0.296, Tokens/s/device: 44.585, total_weights: 16384, loss: 6.459, lm_loss: 6.459, perplexity: 638.363 I0417 14:07:10.466201 140551586671744 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2569.913, Tokens/s/device: 387365.235, total_weights: 16384, loss: 5.622, lm_loss: 5.622, perplexity: 276.577 I0417 14:07:10.584299 140551586671744 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.082, Tokens/s/device: 17497.095, total_weights: 16384, loss: 4.864, lm_loss: 4.864, perplexity: 129.523 I0417 14:07:10.701985 140551586671744 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 14:07:10.702826 140551586671744 checkpointing.py:776] Waited 0.0008530616760253906 seconds for step 9 to finish before starting checkpointing. I0417 14:07:10.703119 140551586671744 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:07:10.947314 140448358204992 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_16_zarr3_false/checkpoints/5/items/array_metadatas/process_0 I0417 14:07:40.877890 140448276416064 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.655224s (commit=30.217978s, array_metadata_write=0.437246s) I0417 14:07:40.878947 140448419022400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 322.223 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0417 14:07:40.879055 140448419022400 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.655571s. I0417 14:07:41.292577 140448419022400 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 14:07:41.820904 140448419022400 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.662642s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5/items I0417 14:07:41.821612 140448419022400 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5/items I0417 14:07:42.058779 140448419022400 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5 I0417 14:07:42.736126 140448419022400 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_16_zarr3_false/checkpoints/5`. I0417 14:07:42.736751 140448419022400 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.170014s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/5 I0417 14:07:42.736826 140448419022400 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.513354s. I0417 14:07:42.737031 140448337233472 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:07:42.737149 140448337233472 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:07:42.737226 140448337233472 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:07:42.737269 140448337233472 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0417 14:07:42.737436 140551586671744 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. W0417 14:07:42.737543 140551586671744 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.034430 seconds. If this number is high, consider checkpointing less frequently. I0417 14:07:42.737775 140551586671744 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 14:07:42.738041 140551586671744 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9. I0417 14:07:42.908156 140551586671744 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 14:07:42.908245 140551586671744 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 14:07:42.911751 140448337233472 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9 I0417 14:07:43.615221 140448297387584 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9/items I0417 14:07:47.660615 140551586671744 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.752876s I0417 14:07:47.663979 140551586671744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.549 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0417 14:07:47.664067 140551586671744 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.840653s (batch_requests_ready=0.081316s, total_serialization_initiated=4.756049s, others=0.003288s) I0417 14:07:47.664137 140551586671744 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.841368s (all_items=0.000014s, per_item={'items': '0.00001407'}, temp_paths=4.841354) I0417 14:07:47.665915 140448419022400 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 14:07:47.666016 140551586671744 async_checkpointer.py:561] Finished blocking save. Time taken: 4.928195s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9. I0417 14:07:47.666253 140551586671744 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 14:07:47.666447 140448347719232 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 14:07:47.666561 140551586671744 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776434830.703094, 'wait_for_prev_duration_secs': 32.0344295501709, 'time_between_consecutive_saves_sec': 9.137499332427979, 'checkpointer_blocking_start_time': 1776434862.7377977, 'checkpointer_blocking_duration_secs': 4.928369522094727, 'get_old_steps_start_time': 1776434867.6661966, 'get_old_steps_duration_secs': 2.1219253540039062e-05, 'checkpoint_manager_blocking_start_time': 1776434830.7030602, 'checkpoint_manager_blocking_duration_secs': 36.96347498893738} I0417 14:07:47.666697 140551586671744 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 14:07:47.666728 140551586671744 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 14:07:48.434587 140448358204992 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_16_zarr3_false/checkpoints/9/items/array_metadatas/process_0 I0417 14:08:18.885551 140448297387584 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.220463s (commit=30.749617s, array_metadata_write=0.470846s) I0417 14:08:18.886774 140448419022400 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 350.406 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host) I0417 14:08:18.886884 140448419022400 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.220801s. I0417 14:08:19.320734 140448419022400 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 14:08:19.884845 140448419022400 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.710187s. use_zarr3=False, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9/items I0417 14:08:19.885542 140448419022400 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9/items I0417 14:08:20.126497 140448419022400 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9 I0417 14:08:20.810981 140448419022400 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_16_zarr3_false/checkpoints/9`. I0417 14:08:20.811637 140448419022400 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.073822s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_16_zarr3_false/checkpoints/9 I0417 14:08:20.811704 140448419022400 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.145622s. I0417 14:08:20.811845 140448347719232 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 14:08:20.811946 140448347719232 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 14:08:20.812014 140448347719232 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 14:08:20.812064 140448347719232 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 14:08:20.812205 140551586671744 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 14:08:20.812353 140551586671744 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 14:08:20.814191 140551586671744 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.716, Tokens/s/device: 17291.310, 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