2026-04-17 13:21:21.439450: 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 13:21:21.556767 137758071237760 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:22:06.344509 137758071237760 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:22:06.344620 137758071237760 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:22:06.344655 137758071237760 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 13:22:06.344680 137758071237760 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:22:06.344763 137758071237760 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:22:06.440060 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:22:06.824270 137758071237760 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:22:06.824683 137758071237760 checkpointing.py:677] Setting up checkpoint logger... I0417 13:22:06.824737 137758071237760 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:22:06.824829 137758071237760 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:22:06.825920 137758071237760 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 0x7d49b5ff8740>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:22:09.232747 137758071237760 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:22:09.232974 137758071237760 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 0x7d433a1421b0>}, handler_registry=None I0417 13:22:09.233363 137758071237760 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d433a1421b0>` 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 13:22:09.233408 137758071237760 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d4339d3a480>` 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 13:22:09.233439 137758071237760 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 0x7d433a1421b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d433a1421b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d4339d3a480>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d4339d3a480>}). I0417 13:22:09.233754 137758071237760 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:22:09.233812 137758071237760 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7d4339d2a200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:22:09.361859 137758071237760 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints I0417 13:22:09.362139 137758071237760 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_10_shardy_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d4339d38e00> I0417 13:22:09.362232 137758071237760 checkpointing.py:301] Checkpoint manager created! I0417 13:22:09.815914 137758071237760 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:22:09.818762 137758071237760 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0417 13:22:09.871513 137758071237760 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0417 13:22:09.894782 137758071237760 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:22:09.894845 137758071237760 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0417 13:22:10.501037 137758071237760 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:22:10.501246 137758071237760 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:22:10.604233 137758071237760 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:22:10.604310 137758071237760 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:22:10.619236 137758071237760 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:22:10.619285 137758071237760 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:22:10.641350 137758071237760 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 13:22:10.641410 137758071237760 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:22:10.656363 137758071237760 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 13:22:10.656413 137758071237760 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:22:10.671568 137758071237760 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 13:22:10.671617 137758071237760 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:22:10.694039 137758071237760 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:22:10.694118 137758071237760 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:22:10.713169 137758071237760 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:22:10.713221 137758071237760 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:22:10.909508 137758071237760 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:22:10.909597 137758071237760 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 13:22:10.960033 454835 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 13:22:11.377224 137758071237760 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:22:11.377348 137758071237760 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 13:22:11.377411 137758071237760 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 13:22:11.377457 137758071237760 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 13:22:11.377496 137758071237760 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 13:22:11.377523 137758071237760 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 13:22:11.377561 137758071237760 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 13:22:11.377614 137758071237760 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 13:22:11.377643 137758071237760 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 13:22:11.377668 137758071237760 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 13:22:11.377704 137758071237760 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:22:11.377738 137758071237760 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:22:11.819725 137758071237760 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:22:11.819806 137758071237760 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:22:11.833957 137758071237760 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:22:11.834007 137758071237760 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:22:12.105486 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:22:12.213869 137758071237760 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 13:22:12.214394 137758071237760 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:22:12.214551 137758071237760 metric_logger.py:301] number parameters: 1.104 billion W0417 13:22:15.258040 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:22:15.388890 137758071237760 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:22:15.508814 137758071237760 checkpointing.py:776] Waited 0.11991381645202637 seconds for step 0 to finish before starting checkpointing. I0417 13:22:15.509369 137758071237760 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:22:15.509538 137758071237760 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:22:15.509940 137758071237760 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0. I0417 13:22:15.602257 137758071237760 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:22:16.080551 137644874925632 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0 I0417 13:22:16.086949 137758071237760 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:22:16.087035 137758071237760 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 13:22:16.302907 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.321823 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.332188 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.344238 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.356396 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.362664 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.367829 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.372492 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:22:16.794357 137644864439872 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/items I0417 13:22:16.802235 137758071237760 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.715763s I0417 13:22:16.802606 137758071237760 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 10.292 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0417 13:22:16.802664 137758071237760 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.199148s (batch_requests_ready=0.479620s, total_serialization_initiated=0.719246s, others=0.000281s) I0417 13:22:16.802724 137758071237760 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.200008s (all_items=0.000049s, per_item={'items': '0.00004888'}, temp_paths=1.199960) I0417 13:22:16.803497 137644803622464 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:22:16.803594 137758071237760 async_checkpointer.py:561] Finished blocking save. Time taken: 1.294014s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0. I0417 13:22:16.803780 137758071237760 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:22:16.803971 137758071237760 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432135.5093539, 'wait_for_prev_duration_secs': 4.482269287109375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432135.5095599, 'checkpointer_blocking_duration_secs': 1.2941455841064453, 'get_old_steps_start_time': 1776432136.8037271, 'get_old_steps_duration_secs': 1.9550323486328125e-05, 'checkpoint_manager_blocking_start_time': 1776432135.5092206, 'checkpoint_manager_blocking_duration_secs': 1.2947282791137695} I0417 13:22:16.804056 137758071237760 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:22:16.804100 137758071237760 max_utils.py:750] Memstats: After params initialized: I0417 13:22:16.804181 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:22:16.804207 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:22:16.804226 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:22:16.804243 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:22:16.804260 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:22:16.804276 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:22:16.804292 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:22:16.804308 137758071237760 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:22:16.810538 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.815891 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:22:16.820579 454835 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:22:16.954259 137644824593984 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432136709986156, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776432136709986156, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:22:16.954508 137644885411392 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:22:17.142409 137758071237760 metric_logger.py:196] completed step: 0, seconds: 3.174, TFLOP/s/device: 4.281, Tokens/s/device: 645.281, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0417 13:22:17.143277 137758071237760 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/tensorboard/' I0417 13:22:17.258036 137758071237760 metric_logger.py:196] completed step: 1, seconds: 1.751, TFLOP/s/device: 7.758, Tokens/s/device: 1169.340, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0417 13:22:17.375352 137758071237760 metric_logger.py:196] completed step: 2, seconds: 0.009, TFLOP/s/device: 1478.308, Tokens/s/device: 222826.678, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0417 13:22:17.485524 458022 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:22:17.485583 458022 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:22:17.493276 137758071237760 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.578, Tokens/s/device: 17571.858, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0417 13:22:18.417608 137644874925632 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_10_shardy_true/checkpoints/0/items/array_metadatas/process_0 I0417 13:22:37.240960 137758071237760 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.589, Tokens/s/device: 17422.818, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0417 13:22:37.250836 137758071237760 metric_logger.py:196] completed step: 5, seconds: 0.133, TFLOP/s/device: 102.097, Tokens/s/device: 15389.240, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0417 13:22:37.367345 137758071237760 metric_logger.py:196] completed step: 6, seconds: 19.733, TFLOP/s/device: 0.689, Tokens/s/device: 103.786, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0417 13:22:37.484979 137758071237760 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 2176.727, Tokens/s/device: 328099.968, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0417 13:22:37.602401 137758071237760 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.816, Tokens/s/device: 17456.976, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0417 13:22:37.719500 137758071237760 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:22:37.720240 137758071237760 checkpointing.py:776] Waited 0.0007584095001220703 seconds for step 9 to finish before starting checkpointing. I0417 13:22:37.720537 137758071237760 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:22:41.072691 137644814108224 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 23.591373s (commit=23.138995s, array_metadata_write=0.452378s) I0417 13:22:41.073737 137644803622464 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 496.140 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host) I0417 13:22:41.073804 137644803622464 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.270160s. I0417 13:22:41.353184 137644803622464 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432136709986156, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:22:41.558228 137644803622464 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 13:22:41.753220 137644824593984 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432136709986156, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:22:41.978442 137644803622464 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:22:41.979105 137644803622464 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.573797s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/items I0417 13:22:41.979743 137644803622464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0/items I0417 13:22:42.227773 137644803622464 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0 I0417 13:22:42.912543 137644803622464 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_10_shardy_true/checkpoints/0`. I0417 13:22:42.913192 137644803622464 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 27.403614s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/0 I0417 13:22:42.913271 137644803622464 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 26.109632s. I0417 13:22:42.913475 137644885411392 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:22:42.913583 137644885411392 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:22:42.913654 137644885411392 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:22:42.913709 137644885411392 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:22:42.913840 137758071237760 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 13:22:42.913969 137758071237760 checkpoint_manager.py:1441] Waiting for previous save to complete took 5.193455 seconds. If this number is high, consider checkpointing less frequently. I0417 13:22:42.915202 137758071237760 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:22:42.915496 137758071237760 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9. I0417 13:22:43.079681 137644885411392 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9 I0417 13:22:43.120143 137758071237760 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:22:43.120240 137758071237760 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 13:22:43.765480 137644692473408 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9/items I0417 13:22:49.201774 137758071237760 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.082006s I0417 13:22:49.206160 137758071237760 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.989 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0417 13:22:49.206219 137758071237760 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.205396s (batch_requests_ready=0.115972s, total_serialization_initiated=6.085148s, others=0.004276s) I0417 13:22:49.206291 137758071237760 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.205946s (all_items=0.000025s, per_item={'items': '0.00002480'}, temp_paths=6.205922) I0417 13:22:49.207819 137644885411392 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:22:49.207903 137758071237760 async_checkpointer.py:561] Finished blocking save. Time taken: 6.292655s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9. I0417 13:22:49.208126 137758071237760 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:22:49.208318 137644803622464 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:22:49.208432 137758071237760 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432157.7204816, 'wait_for_prev_duration_secs': 5.193454742431641, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432162.9152277, 'checkpointer_blocking_duration_secs': 6.292778491973877, 'get_old_steps_start_time': 1776432169.20803, 'get_old_steps_duration_secs': 5.53131103515625e-05, 'checkpoint_manager_blocking_start_time': 1776432157.7204263, 'checkpoint_manager_blocking_duration_secs': 11.48798131942749} I0417 13:22:49.208554 137758071237760 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:22:49.208588 137758071237760 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:22:49.947313 137644874925632 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_10_shardy_true/checkpoints/9/items/array_metadatas/process_0 I0417 13:23:20.668724 137644692473408 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.461984s (commit=31.020324s, array_metadata_write=0.441660s) I0417 13:23:20.669710 137644885411392 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 335.470 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0417 13:23:20.669771 137644885411392 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.461820s. I0417 13:23:21.080790 137644885411392 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 13:23:21.498499 137644885411392 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:23:21.498677 137644885411392 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.559712s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9/items I0417 13:23:21.499327 137644885411392 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9/items I0417 13:23:21.732209 137644885411392 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9 I0417 13:23:22.382137 137644885411392 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_10_shardy_true/checkpoints/9`. I0417 13:23:22.382775 137644885411392 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.467529s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_10_shardy_true/checkpoints/9 I0417 13:23:22.382854 137644885411392 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.174907s. I0417 13:23:22.383039 137644803622464 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:23:22.383147 137644803622464 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:23:22.383233 137644803622464 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:23:22.383282 137644803622464 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:23:22.383465 137758071237760 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 13:23:22.383903 137758071237760 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:23:22.384668 137758071237760 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.369, Tokens/s/device: 17389.680, total_weights: 12300, loss: 8.710, lm_loss: 8.710, perplexity: 6065.008 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops