2026-04-17 13:26:36.676502: 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 13:26:36.795308 128182749510784 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:27:25.530666 128182749510784 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:27:25.530779 128182749510784 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:27:25.530822 128182749510784 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 13:27:25.530848 128182749510784 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:27:25.530875 128182749510784 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:27:25.530966 128182749510784 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:27:25.627497 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:26.028953 128182749510784 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:27:26.029323 128182749510784 checkpointing.py:677] Setting up checkpoint logger... I0417 13:27:26.029368 128182749510784 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:27:26.029444 128182749510784 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:27:26.030101 128182749510784 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 0x7494485f8b30>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:27:28.468714 128182749510784 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:27:28.468943 128182749510784 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 0x748fc157a960>}, handler_registry=None I0417 13:27:28.469332 128182749510784 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x748fc157a960>` 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 13:27:28.469377 128182749510784 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748dcc331610>` 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 13:27:28.469409 128182749510784 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 0x748fc157a960>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x748fc157a960>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748dcc331610>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x748dcc331610>}). I0417 13:27:28.470055 128182749510784 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:27:28.470117 128182749510784 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x748dcc32df80> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:27:28.596027 128182749510784 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints I0417 13:27:28.596292 128182749510784 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x748dcc331850> I0417 13:27:28.596380 128182749510784 checkpointing.py:301] Checkpoint manager created! W0417 13:27:28.613095 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:29.139616 128182749510784 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:27:29.139714 128182749510784 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:27:29.222704 128182749510784 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:27:29.222785 128182749510784 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:27:29.237009 128182749510784 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:27:29.237068 128182749510784 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:27:29.257279 128182749510784 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 13:27:29.257332 128182749510784 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:27:29.271532 128182749510784 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 13:27:29.271579 128182749510784 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:27:29.285735 128182749510784 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 13:27:29.285783 128182749510784 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:27:29.307070 128182749510784 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:27:29.307125 128182749510784 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:27:29.325257 128182749510784 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:27:29.325308 128182749510784 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:27:29.518222 128182749510784 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:27:29.518311 128182749510784 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 13:27:29.567535 464214 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 13:27:29.947059 128182749510784 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:27:29.947163 128182749510784 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 13:27:29.947203 128182749510784 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 13:27:29.947246 128182749510784 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 13:27:29.947284 128182749510784 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 13:27:29.947312 128182749510784 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 13:27:29.947354 128182749510784 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 13:27:29.947394 128182749510784 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 13:27:29.947423 128182749510784 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 13:27:29.947448 128182749510784 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 13:27:29.947482 128182749510784 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:27:29.947516 128182749510784 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:27:30.365477 128182749510784 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:27:30.365555 128182749510784 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:27:30.379172 128182749510784 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:27:30.379222 128182749510784 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:27:30.631421 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:30.744503 128182749510784 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 13:27:30.744942 128182749510784 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:27:30.745099 128182749510784 metric_logger.py:301] number parameters: 1.104 billion W0417 13:27:31.597695 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:31.694458 128182749510784 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:27:31.811402 128182749510784 checkpointing.py:776] Waited 0.11696410179138184 seconds for step 0 to finish before starting checkpointing. I0417 13:27:31.812015 128182749510784 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:27:31.812211 128182749510784 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:27:31.812801 128182749510784 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0. I0417 13:27:31.896779 128182749510784 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:27:31.990781 128182749510784 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:27:31.990871 128182749510784 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 13:27:31.997211 128075264165440 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0 W0417 13:27:32.206782 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.218596 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.234076 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.244239 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.254993 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.262544 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.268137 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.274732 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:32.695446 128075253679680 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/items I0417 13:27:32.735146 128182749510784 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.744822s I0417 13:27:32.735560 128182749510784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 14.731 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 837 milliseconds) (per-host) I0417 13:27:32.735614 128182749510784 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.837830s (batch_requests_ready=0.089107s, total_serialization_initiated=0.748396s, others=0.000327s) I0417 13:27:32.735670 128182749510784 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.838407s (all_items=0.000031s, per_item={'items': '0.00003076'}, temp_paths=0.838376) I0417 13:27:32.736376 128075197056576 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:27:32.736452 128182749510784 async_checkpointer.py:561] Finished blocking save. Time taken: 0.924198s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0. I0417 13:27:32.736649 128182749510784 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:27:32.736833 128182749510784 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432451.811998, 'wait_for_prev_duration_secs': 5.7697296142578125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432451.8122327, 'checkpointer_blocking_duration_secs': 0.9243214130401611, 'get_old_steps_start_time': 1776432452.7365768, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776432451.811852, 'checkpoint_manager_blocking_duration_secs': 0.9249577522277832} I0417 13:27:32.736903 128182749510784 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:27:32.736946 128182749510784 max_utils.py:750] Memstats: After params initialized: I0417 13:27:32.737017 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:27:32.737061 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:27:32.737083 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:27:32.737102 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:27:32.737121 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:27:32.737138 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:27:32.737155 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:27:32.737172 128182749510784 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:27:32.742746 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.747934 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:27:32.752586 464214 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:27:32.872751 128075218028096 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432452616939724, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776432452616939724, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 13:27:32.872996 128075278845504 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:27:33.071194 128182749510784 metric_logger.py:196] completed step: 0, seconds: 0.949, TFLOP/s/device: 14.320, Tokens/s/device: 2158.395, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 13:27:33.072228 128182749510784 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/tensorboard/' I0417 13:27:33.187465 128182749510784 metric_logger.py:196] completed step: 1, seconds: 1.375, TFLOP/s/device: 9.884, Tokens/s/device: 1489.847, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 13:27:33.306645 128182749510784 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 1964.310, Tokens/s/device: 296082.117, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0417 13:27:33.386433 466832 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:27:33.386486 466832 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:27:33.424795 128182749510784 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.139, Tokens/s/device: 17505.770, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0417 13:27:34.170699 128075264165440 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_0 I0417 13:27:53.190790 128182749510784 metric_logger.py:196] completed step: 4, seconds: 0.119, TFLOP/s/device: 114.074, Tokens/s/device: 17194.479, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0417 13:27:53.198823 128182749510784 metric_logger.py:196] completed step: 5, seconds: 0.118, TFLOP/s/device: 114.881, Tokens/s/device: 17316.164, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0417 13:27:53.315794 128182749510784 metric_logger.py:196] completed step: 6, seconds: 19.767, TFLOP/s/device: 0.687, Tokens/s/device: 103.606, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0417 13:27:53.433915 128182749510784 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2709.838, Tokens/s/device: 408456.322, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0417 13:27:53.552096 128182749510784 metric_logger.py:196] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.439, Tokens/s/device: 17400.170, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0417 13:27:53.669911 128182749510784 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:27:53.670723 128182749510784 checkpointing.py:776] Waited 0.0008382797241210938 seconds for step 9 to finish before starting checkpointing. I0417 13:27:53.671007 128182749510784 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:27:59.407826 128075207542336 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.020413s (commit=25.587059s, array_metadata_write=0.433353s) I0417 13:27:59.408849 128075197056576 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 459.335 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 27 seconds) (per-host) I0417 13:27:59.408910 128075197056576 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.672412s. I0417 13:27:59.625832 128075197056576 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432452616939724, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 13:27:59.814208 128075197056576 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 13:28:00.034121 128075218028096 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432452616939724, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/_CHECKPOINT_METADATA I0417 13:28:00.249015 128075197056576 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:28:00.249662 128075197056576 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.576975s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/items I0417 13:28:00.250316 128075197056576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0/items I0417 13:28:00.491374 128075197056576 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0 I0417 13:28:01.185958 128075197056576 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0`. I0417 13:28:01.186577 128075197056576 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 29.374325s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/0 I0417 13:28:01.186665 128075197056576 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.450171s. I0417 13:28:01.186831 128075278845504 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:28:01.186935 128075278845504 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:28:01.187001 128075278845504 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:28:01.187056 128075278845504 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:28:01.187149 128182749510784 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 13:28:01.187230 128182749510784 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.516244 seconds. If this number is high, consider checkpointing less frequently. I0417 13:28:01.187937 128182749510784 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:28:01.188193 128182749510784 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9. I0417 13:28:01.375367 128182749510784 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:28:01.375505 128075278845504 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9 I0417 13:28:01.375584 128182749510784 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 13:28:02.087635 128075043964480 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9/items I0417 13:28:04.348747 128182749510784 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.974099s I0417 13:28:04.349253 128182749510784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.018 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0417 13:28:04.349402 128182749510784 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.071228s (batch_requests_ready=0.093538s, total_serialization_initiated=2.977290s, others=0.000401s) I0417 13:28:04.349479 128182749510784 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.071761s (all_items=0.000022s, per_item={'items': '0.00002217'}, temp_paths=3.071739) I0417 13:28:04.354850 128075207542336 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:28:04.354970 128182749510784 async_checkpointer.py:561] Finished blocking save. Time taken: 3.166980s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9. I0417 13:28:04.355321 128182749510784 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:28:04.355623 128075197056576 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:28:04.355741 128182749510784 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432473.6709619, 'wait_for_prev_duration_secs': 7.516244173049927, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432481.1879609, 'checkpointer_blocking_duration_secs': 3.1671924591064453, 'get_old_steps_start_time': 1776432484.355208, 'get_old_steps_duration_secs': 4.9591064453125e-05, 'checkpoint_manager_blocking_start_time': 1776432473.6709137, 'checkpoint_manager_blocking_duration_secs': 10.684800148010254} I0417 13:28:04.355868 128182749510784 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:28:04.355904 128182749510784 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:28:05.091441 128075264165440 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_0 I0417 13:28:48.441117 128075043964480 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 44.087040s (commit=43.640573s, array_metadata_write=0.446467s) I0417 13:28:48.442298 128075207542336 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 267.933 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host) I0417 13:28:48.442409 128075207542336 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.087375s. I0417 13:28:48.834038 128075207542336 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 13:28:49.266113 128075207542336 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:28:49.266834 128075207542336 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.568252s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9/items I0417 13:28:49.267424 128075207542336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9/items I0417 13:28:49.527100 128075207542336 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9 I0417 13:28:50.274094 128075207542336 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9`. I0417 13:28:50.274699 128075207542336 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 49.086719s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_11_optimizer_offload_false/checkpoints/9 I0417 13:28:50.274784 128075207542336 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 45.919757s. I0417 13:28:50.275002 128075197056576 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:28:50.275125 128075197056576 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:28:50.275231 128075197056576 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:28:50.275283 128075197056576 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:28:50.275742 128182749510784 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 13:28:50.276447 128182749510784 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:28:50.277478 128182749510784 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.116, Tokens/s/device: 17351.521, total_weights: 16384, loss: 5.915, lm_loss: 5.915, perplexity: 370.653 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops