2026-04-16 00:51:05.077882: 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) I0416 00:51:05.193565 130542889118848 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 00:51:36.830867 130542889118848 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 00:51:36.830981 130542889118848 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 00:51:36.831016 130542889118848 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 I0416 00:51:36.831107 130542889118848 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 00:51:36.831131 130542889118848 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0416 00:51:36.831212 130542889118848 train.py:812] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 00:51:36.924157 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:37.339918 130542889118848 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 00:51:37.340113 130542889118848 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 00:51:37.340294 130542889118848 checkpointing.py:688] Setting up checkpoint logger... I0416 00:51:37.340334 130542889118848 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 00:51:37.340380 130542889118848 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 00:51:37.340817 130542889118848 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 0x76b9cc3f8da0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 00:51:39.762161 130542889118848 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 00:51:39.762357 130542889118848 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 0x76b34fb31c70>}, handler_registry=None I0416 00:51:39.762621 130542889118848 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76b34fb31c70>` 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`. I0416 00:51:39.762661 130542889118848 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76b350904f50>` 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`. I0416 00:51:39.762689 130542889118848 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 0x76b34fb31c70>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76b34fb31c70>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76b350904f50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76b350904f50>}). I0416 00:51:39.762972 130542889118848 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 00:51:39.763039 130542889118848 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x76b34fb3e0c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 00:51:39.888545 130542889118848 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints I0416 00:51:39.888794 130542889118848 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_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x76b34ff16cc0> I0416 00:51:39.888878 130542889118848 checkpointing.py:302] Checkpoint manager created! W0416 00:51:39.905910 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:40.099239 130542889118848 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 00:51:40.099337 130542889118848 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:51:40.189776 130542889118848 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 00:51:40.189862 130542889118848 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:51:40.216638 130542889118848 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 00:51:40.216762 130542889118848 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 00:51:40.239403 130542889118848 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:51:40.239469 130542889118848 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:51:40.253595 130542889118848 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:51:40.253644 130542889118848 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:51:40.268039 130542889118848 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 00:51:40.268101 130542889118848 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:51:40.290017 130542889118848 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 00:51:40.290090 130542889118848 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 00:51:40.307956 130542889118848 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 00:51:40.308014 130542889118848 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 00:51:40.499257 130542889118848 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 00:51:40.499370 130542889118848 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 00:51:40.552082 3446523 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 I0416 00:51:40.930692 130542889118848 maxtext_utils.py:1790] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 00:51:40.930807 130542889118848 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 00:51:40.930850 130542889118848 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 00:51:40.930896 130542889118848 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0416 00:51:40.930940 130542889118848 maxtext_utils.py:1790] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 00:51:40.930976 130542889118848 maxtext_utils.py:1790] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 00:51:40.931016 130542889118848 maxtext_utils.py:1790] 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) I0416 00:51:40.931072 130542889118848 maxtext_utils.py:1790] 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') I0416 00:51:40.931104 130542889118848 maxtext_utils.py:1790] 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) I0416 00:51:40.931130 130542889118848 maxtext_utils.py:1790] 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) I0416 00:51:40.931165 130542889118848 maxtext_utils.py:1790] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 00:51:40.931198 130542889118848 maxtext_utils.py:1790] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 00:51:41.355098 130542889118848 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 00:51:41.355187 130542889118848 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 00:51:41.369886 130542889118848 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 00:51:41.369940 130542889118848 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 00:51:42.047608 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:42.166827 130542889118848 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. I0416 00:51:42.167349 130542889118848 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 00:51:42.167493 130542889118848 metric_logger.py:289] number parameters: 1.104 billion W0416 00:51:43.052626 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:43.150221 130542889118848 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 00:51:43.267515 130542889118848 checkpointing.py:798] Waited 0.11727523803710938 seconds for step 0 to finish before starting checkpointing. I0416 00:51:43.268021 130542889118848 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 00:51:43.268226 130542889118848 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 00:51:43.268648 130542889118848 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0. I0416 00:51:43.359396 130542889118848 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 00:51:43.444905 130435409512000 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0 I0416 00:51:43.454592 130542889118848 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 00:51:43.454694 130542889118848 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 W0416 00:51:43.679785 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.693636 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.707355 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.715637 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.729202 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.736031 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.742545 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.749699 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:43.935511 130542889118848 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.481373s I0416 00:51:43.935869 130542889118848 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 21.450 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 575 milliseconds) (per-host) I0416 00:51:43.935918 130542889118848 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.575431s (batch_requests_ready=0.089849s, total_serialization_initiated=0.485345s, others=0.000237s) I0416 00:51:43.935986 130542889118848 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.576032s (all_items=0.000021s, per_item={'items': '0.00002098'}, temp_paths=0.576011) I0416 00:51:43.936899 130435346597440 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 00:51:43.937001 130542889118848 async_checkpointer.py:561] Finished blocking save. Time taken: 0.668734s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0. I0416 00:51:43.937193 130542889118848 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 00:51:43.937363 130435419997760 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 00:51:43.937446 130542889118848 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776300703.2680066, 'wait_for_prev_duration_secs': 8.225440979003906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776300703.2682474, 'checkpointer_blocking_duration_secs': 0.6688649654388428, 'get_old_steps_start_time': 1776300703.9371347, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776300703.2679102, 'checkpoint_manager_blocking_duration_secs': 0.6695032119750977} I0416 00:51:43.937563 130542889118848 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 00:51:43.937619 130542889118848 max_utils.py:750] Memstats: After params initialized: I0416 00:51:43.937705 130542889118848 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0416 00:51:43.937732 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 00:51:43.937751 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 00:51:43.937773 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 00:51:43.937789 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 00:51:43.937805 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 00:51:43.937821 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 00:51:43.937837 130542889118848 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0416 00:51:43.943823 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.949250 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 00:51:43.953998 3446523 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 00:51:44.106830 130435399026240 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/items I0416 00:51:44.264229 130435327723072 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776300704024772887, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776300704024772887, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 00:51:44.270974 130542889118848 metric_logger.py:185] completed step: 0, seconds: 0.982, TFLOP/s/device: 13.833, Tokens/s/device: 2085.005, total_weights: 16384, loss: 10.874 I0416 00:51:44.271930 130542889118848 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/tensorboard/' I0416 00:51:44.387213 130542889118848 metric_logger.py:185] completed step: 1, seconds: 1.119, TFLOP/s/device: 12.143, Tokens/s/device: 1830.397, total_weights: 16384, loss: 10.874 I0416 00:51:44.505986 130542889118848 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 2050.268, Tokens/s/device: 309038.781, total_weights: 16384, loss: 9.775 I0416 00:51:44.624686 130542889118848 metric_logger.py:185] completed step: 3, seconds: 0.127, TFLOP/s/device: 107.164, Tokens/s/device: 16152.948, total_weights: 16384, loss: 8.838 I0416 00:51:44.660407 3448830 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 00:51:44.660607 3448830 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 00:51:45.440736 130435367568960 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/items/array_metadatas/process_0 I0416 00:52:04.406679 130542889118848 metric_logger.py:185] completed step: 4, seconds: 0.109, TFLOP/s/device: 124.095, Tokens/s/device: 18704.905, total_weights: 16384, loss: 7.965 I0416 00:52:04.415272 130542889118848 metric_logger.py:185] completed step: 5, seconds: 0.118, TFLOP/s/device: 114.693, Tokens/s/device: 17287.807, total_weights: 16384, loss: 7.233 I0416 00:52:04.531705 130542889118848 metric_logger.py:185] completed step: 6, seconds: 19.784, TFLOP/s/device: 0.687, Tokens/s/device: 103.521, total_weights: 16384, loss: 6.676 I0416 00:52:04.650180 130542889118848 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 2625.025, Tokens/s/device: 395672.334, total_weights: 16384, loss: 6.293 I0416 00:52:04.768589 130542889118848 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.831, Tokens/s/device: 17459.357, total_weights: 16384, loss: 6.055 I0416 00:52:04.885915 130542889118848 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 00:52:04.886846 130542889118848 checkpointing.py:798] Waited 0.0009629726409912109 seconds for step 9 to finish before starting checkpointing. I0416 00:52:04.887176 130542889118848 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 00:52:07.823784 130435357083200 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 23.170711s (commit=22.754117s, array_metadata_write=0.416594s) I0416 00:52:07.824955 130435346597440 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 516.537 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 24 seconds) (per-host) I0416 00:52:07.825021 130435346597440 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 23.887969s. I0416 00:52:08.047976 130435346597440 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776300704024772887, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 00:52:08.229865 130435346597440 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 00:52:08.445166 130435327723072 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776300704024772887, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/_CHECKPOINT_METADATA I0416 00:52:08.644621 130435346597440 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 00:52:08.644820 130435346597440 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.550998s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/items I0416 00:52:08.645984 130435346597440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0/items I0416 00:52:08.892037 130435346597440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0 I0416 00:52:09.573846 130435346597440 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0`. I0416 00:52:09.574635 130435346597440 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 26.306372s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/0 I0416 00:52:09.574713 130435346597440 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 25.637668s. I0416 00:52:09.574840 130435419997760 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 00:52:09.574896 130435419997760 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 00:52:09.574953 130435419997760 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 00:52:09.574996 130435419997760 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 00:52:09.575182 130542889118848 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. W0416 00:52:09.575333 130542889118848 checkpoint_manager.py:1441] Waiting for previous save to complete took 4.688174 seconds. If this number is high, consider checkpointing less frequently. I0416 00:52:09.576602 130542889118848 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 00:52:09.576934 130542889118848 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9. I0416 00:52:09.736097 130542889118848 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 00:52:09.736197 130542889118848 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 I0416 00:52:09.792015 130435419997760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9 I0416 00:52:10.479704 130435399026240 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9/items I0416 00:52:21.226596 130542889118848 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.490997s I0416 00:52:21.227191 130542889118848 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.066 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0416 00:52:21.227244 130542889118848 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.573772s (batch_requests_ready=0.079068s, total_serialization_initiated=11.494242s, others=0.000462s) I0416 00:52:21.227307 130542889118848 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.574361s (all_items=0.000024s, per_item={'items': '0.00002360'}, temp_paths=11.574338) I0416 00:52:21.232700 130435346597440 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 00:52:21.232814 130542889118848 async_checkpointer.py:561] Finished blocking save. Time taken: 11.656165s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9. I0416 00:52:21.233035 130542889118848 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 00:52:21.233240 130435357083200 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 00:52:21.233364 130542889118848 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776300724.887123, 'wait_for_prev_duration_secs': 4.68817400932312, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776300729.5766256, 'checkpointer_blocking_duration_secs': 11.656302690505981, 'get_old_steps_start_time': 1776300741.2329478, 'get_old_steps_duration_secs': 4.458427429199219e-05, 'checkpoint_manager_blocking_start_time': 1776300724.8870401, 'checkpoint_manager_blocking_duration_secs': 16.34629535675049} I0416 00:52:21.233487 130542889118848 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 00:52:21.233520 130542889118848 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 00:52:21.975746 130435388540480 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9/items/array_metadatas/process_0 I0416 00:52:53.943462 130435399026240 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.711663s (commit=32.308826s, array_metadata_write=0.402837s) I0416 00:52:53.944660 130435346597440 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 285.312 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0416 00:52:53.944732 130435346597440 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.711867s. I0416 00:52:54.350111 130435346597440 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 00:52:54.775643 130435346597440 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 00:52:54.775842 130435346597440 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.568562s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9/items I0416 00:52:54.776563 130435346597440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9/items I0416 00:52:55.003678 130435346597440 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9 I0416 00:52:55.700829 130435346597440 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9`. I0416 00:52:55.701594 130435346597440 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.124950s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_004836/linen_feat_nnx_trainstate_and_training_loop_20260416_004836_02_synthetic/checkpoints/9 I0416 00:52:55.701672 130435346597440 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.468812s. I0416 00:52:55.701892 130435357083200 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 00:52:55.702023 130435357083200 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 00:52:55.702121 130435357083200 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 00:52:55.702170 130435357083200 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 00:52:55.702459 130542889118848 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. I0416 00:52:55.702608 130542889118848 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 00:52:55.703285 130542889118848 metric_logger.py:185] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.686, Tokens/s/device: 17286.785, total_weights: 16384, loss: 5.915 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops