2026-04-17 13:36:32.043063: 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 13:36:32.161006 131006703238272 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0417 13:37:19.512680 131006703238272 max_utils.py:800] System Information: Jax Version: 0.9.2 I0417 13:37:19.512795 131006703238272 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0417 13:37:19.512833 131006703238272 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 13:37:19.512861 131006703238272 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0417 13:37:19.512884 131006703238272 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0417 13:37:19.513006 131006703238272 train.py:671] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0417 13:37:19.606308 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:20.009191 131006703238272 maxtext_utils.py:1548] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0417 13:37:20.009546 131006703238272 checkpointing.py:677] Setting up checkpoint logger... I0417 13:37:20.009592 131006703238272 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0417 13:37:20.009654 131006703238272 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0417 13:37:20.010333 131006703238272 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 0x7725c9103c50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0417 13:37:22.398822 131006703238272 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0417 13:37:22.399014 131006703238272 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 0x771f4d736390>}, handler_registry=None I0417 13:37:22.399327 131006703238272 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x771f4d736390>` 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 13:37:22.399369 131006703238272 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x771f4d735250>` 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 13:37:22.399397 131006703238272 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 0x771f4d736390>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x771f4d736390>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x771f4d735250>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x771f4d735250>}). I0417 13:37:22.399688 131006703238272 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0417 13:37:22.399743 131006703238272 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x771f4d73e160> timeout: 600 secs and primary_host=0 for async checkpoint writes I0417 13:37:22.525173 131006703238272 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints I0417 13:37:22.525402 131006703238272 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_13_scan_layers_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x771f71178ce0> I0417 13:37:22.525485 131006703238272 checkpointing.py:301] Checkpoint manager created! W0417 13:37:22.542154 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:23.083674 131006703238272 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0417 13:37:23.083773 131006703238272 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:37:23.167008 131006703238272 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:37:23.167101 131006703238272 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:37:23.181341 131006703238272 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0417 13:37:23.181390 131006703238272 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0417 13:37:23.201882 131006703238272 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 13:37:23.201938 131006703238272 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:37:23.216109 131006703238272 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 13:37:23.216157 131006703238272 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:37:23.230299 131006703238272 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 13:37:23.230345 131006703238272 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:37:23.251786 131006703238272 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0417 13:37:23.251844 131006703238272 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0417 13:37:23.269913 131006703238272 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0417 13:37:23.269963 131006703238272 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0417 13:37:23.458208 131006703238272 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0417 13:37:23.458303 131006703238272 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0417 13:37:23.507644 477931 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 13:37:23.883790 131006703238272 maxtext_utils.py:1651] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0417 13:37:23.883899 131006703238272 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 13:37:23.883947 131006703238272 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 13:37:23.883990 131006703238272 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 13:37:23.884029 131006703238272 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 13:37:23.884091 131006703238272 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 13:37:23.884135 131006703238272 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 13:37:23.884174 131006703238272 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 13:37:23.884202 131006703238272 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 13:37:23.884227 131006703238272 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 13:37:23.884261 131006703238272 maxtext_utils.py:1651] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0417 13:37:23.884295 131006703238272 maxtext_utils.py:1651] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0417 13:37:24.301544 131006703238272 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:37:24.301626 131006703238272 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0417 13:37:24.315114 131006703238272 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0417 13:37:24.315162 131006703238272 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0417 13:37:24.563421 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:24.685497 131006703238272 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 13:37:24.685973 131006703238272 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0417 13:37:24.686125 131006703238272 metric_logger.py:301] number parameters: 1.104 billion W0417 13:37:25.516794 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:25.612630 131006703238272 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0417 13:37:25.729792 131006703238272 checkpointing.py:776] Waited 0.11715316772460938 seconds for step 0 to finish before starting checkpointing. I0417 13:37:25.730249 131006703238272 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:37:25.730395 131006703238272 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0417 13:37:25.730749 131006703238272 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0. I0417 13:37:25.814038 131006703238272 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0417 13:37:25.900227 130903521101376 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0 I0417 13:37:25.908013 131006703238272 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:37:25.908116 131006703238272 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 13:37:26.121991 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.135312 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.154338 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.171509 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.178679 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.184943 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.189945 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.194681 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:26.502650 131006703238272 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.595051s I0417 13:37:26.503203 131006703238272 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 17.943 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 687 milliseconds) (per-host) I0417 13:37:26.503350 131006703238272 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.688070s (batch_requests_ready=0.089060s, total_serialization_initiated=0.598540s, others=0.000470s) I0417 13:37:26.503470 131006703238272 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.688962s (all_items=0.000021s, per_item={'items': '0.00002074'}, temp_paths=0.688941) I0417 13:37:26.504330 130903531587136 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:37:26.504419 131006703238272 async_checkpointer.py:561] Finished blocking save. Time taken: 0.773985s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0. I0417 13:37:26.504612 131006703238272 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:37:26.504834 131006703238272 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433045.7302365, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433045.7304144, 'checkpointer_blocking_duration_secs': 0.7741148471832275, 'get_old_steps_start_time': 1776433046.504549, 'get_old_steps_duration_secs': 2.2411346435546875e-05, 'checkpoint_manager_blocking_start_time': 1776433045.7301526, 'checkpoint_manager_blocking_duration_secs': 0.7746546268463135} I0417 13:37:26.504904 131006703238272 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0417 13:37:26.504944 131006703238272 max_utils.py:750] Memstats: After params initialized: I0417 13:37:26.505037 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0417 13:37:26.505083 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0417 13:37:26.505105 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0417 13:37:26.505127 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0417 13:37:26.505145 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0417 13:37:26.505164 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0417 13:37:26.505182 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0417 13:37:26.505199 131006703238272 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0417 13:37:26.510863 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.516100 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0417 13:37:26.520861 477931 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0417 13:37:26.575689 130903510615616 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/items I0417 13:37:26.819002 130903479158336 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433046479034425, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776433046479034425, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:37:26.819695 130903458186816 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:37:26.826039 131006703238272 metric_logger.py:196] completed step: 0, seconds: 0.926, TFLOP/s/device: 14.672, Tokens/s/device: 2211.596, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 13:37:26.827011 131006703238272 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/tensorboard/' I0417 13:37:26.941929 131006703238272 metric_logger.py:196] completed step: 1, seconds: 1.211, TFLOP/s/device: 11.217, Tokens/s/device: 1690.694, total_weights: 16384, loss: 10.874, lm_loss: 10.874, perplexity: 52770.941 I0417 13:37:27.060493 131006703238272 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 1975.448, Tokens/s/device: 297760.977, total_weights: 16384, loss: 9.775, lm_loss: 9.775, perplexity: 17589.287 I0417 13:37:27.178911 131006703238272 metric_logger.py:196] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.626, Tokens/s/device: 17579.097, total_weights: 16384, loss: 8.838, lm_loss: 8.838, perplexity: 6892.933 I0417 13:37:27.225409 480203 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0417 13:37:27.225469 480203 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0417 13:37:28.206124 130903521101376 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_13_scan_layers_true/checkpoints/0/items/array_metadatas/process_0 I0417 13:37:47.353246 131006703238272 metric_logger.py:196] completed step: 4, seconds: 0.118, TFLOP/s/device: 114.675, Tokens/s/device: 17285.034, total_weights: 16384, loss: 7.965, lm_loss: 7.965, perplexity: 2878.644 I0417 13:37:47.361974 131006703238272 metric_logger.py:196] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.656, Tokens/s/device: 17282.263, total_weights: 16384, loss: 7.233, lm_loss: 7.233, perplexity: 1383.750 I0417 13:37:47.478649 131006703238272 metric_logger.py:196] completed step: 6, seconds: 20.176, TFLOP/s/device: 0.673, Tokens/s/device: 101.506, total_weights: 16384, loss: 6.676, lm_loss: 6.676, perplexity: 793.531 I0417 13:37:47.596804 131006703238272 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 2719.057, Tokens/s/device: 409845.908, total_weights: 16384, loss: 6.293, lm_loss: 6.293, perplexity: 540.860 I0417 13:37:47.714942 131006703238272 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.753, Tokens/s/device: 17447.606, total_weights: 16384, loss: 6.055, lm_loss: 6.055, perplexity: 426.371 I0417 13:37:47.832371 131006703238272 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0417 13:37:47.833176 131006703238272 checkpointing.py:776] Waited 0.0008311271667480469 seconds for step 9 to finish before starting checkpointing. I0417 13:37:47.833466 131006703238272 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:38:02.874953 130903468672576 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.653721s (commit=35.231912s, array_metadata_write=0.421809s) I0417 13:38:02.876057 130903531587136 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 340.975 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0417 13:38:02.876183 130903531587136 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.371713s. I0417 13:38:03.098131 130903531587136 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433046479034425, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:38:03.283696 130903531587136 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 13:38:03.489249 130903479158336 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776433046479034425, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/_CHECKPOINT_METADATA I0417 13:38:03.732450 130903531587136 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:38:03.733114 130903531587136 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.583833s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/items I0417 13:38:03.733784 130903531587136 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0/items I0417 13:38:03.976813 130903531587136 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0 I0417 13:38:04.634942 130903531587136 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_13_scan_layers_true/checkpoints/0`. I0417 13:38:04.635573 130903531587136 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.905143s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/0 I0417 13:38:04.635658 130903531587136 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.131194s. I0417 13:38:04.635859 130903458186816 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:38:04.635969 130903458186816 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:38:04.636037 130903458186816 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:38:04.636114 130903458186816 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0417 13:38:04.636248 131006703238272 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 13:38:04.636385 131006703238272 checkpoint_manager.py:1441] Waiting for previous save to complete took 16.802933 seconds. If this number is high, consider checkpointing less frequently. I0417 13:38:04.637476 131006703238272 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0417 13:38:04.637809 131006703238272 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9. I0417 13:38:04.809234 131006703238272 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0417 13:38:04.809328 131006703238272 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 13:38:04.820719 130903458186816 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9 I0417 13:38:05.522190 130903347037760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9/items I0417 13:38:11.998578 131006703238272 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 7.189764s I0417 13:38:12.002885 131006703238272 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.694 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 7 seconds) (per-host) I0417 13:38:12.002948 131006703238272 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 7.284753s (batch_requests_ready=0.087561s, total_serialization_initiated=7.192994s, others=0.004198s) I0417 13:38:12.003019 131006703238272 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 7.285574s (all_items=0.000024s, per_item={'items': '0.00002432'}, temp_paths=7.285549) I0417 13:38:12.004378 130903439312448 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0417 13:38:12.004536 131006703238272 async_checkpointer.py:561] Finished blocking save. Time taken: 7.367012s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9. I0417 13:38:12.004796 131006703238272 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0417 13:38:12.004947 130903531587136 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0417 13:38:12.005015 131006703238272 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776433067.8334138, 'wait_for_prev_duration_secs': 16.80293321609497, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776433084.6375022, 'checkpointer_blocking_duration_secs': 7.367156505584717, 'get_old_steps_start_time': 1776433092.0046773, 'get_old_steps_duration_secs': 3.743171691894531e-05, 'checkpoint_manager_blocking_start_time': 1776433067.8333635, 'checkpoint_manager_blocking_duration_secs': 24.171627044677734} I0417 13:38:12.005136 131006703238272 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0417 13:38:12.005171 131006703238272 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0417 13:38:12.749497 130903521101376 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_13_scan_layers_true/checkpoints/9/items/array_metadatas/process_0 I0417 13:38:38.681612 130903468672576 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.678025s (commit=26.245197s, array_metadata_write=0.432829s) I0417 13:38:38.682683 130903439312448 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 372.059 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0417 13:38:38.682795 130903439312448 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.678215s. I0417 13:38:39.104091 130903439312448 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 13:38:39.532108 130903439312448 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0417 13:38:39.532872 130903439312448 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.568964s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9/items I0417 13:38:39.533543 130903439312448 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9/items I0417 13:38:39.775984 130903439312448 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9 I0417 13:38:40.475032 130903439312448 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_13_scan_layers_true/checkpoints/9`. I0417 13:38:40.475671 130903439312448 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.838153s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260417_125240/linen_main_20260417_125240_13_scan_layers_true/checkpoints/9 I0417 13:38:40.475751 130903439312448 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.471173s. I0417 13:38:40.475857 130903531587136 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0417 13:38:40.475905 130903531587136 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0417 13:38:40.475952 130903531587136 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0417 13:38:40.475991 130903531587136 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0417 13:38:40.476237 131006703238272 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 13:38:40.476365 131006703238272 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0417 13:38:40.477243 131006703238272 metric_logger.py:196] completed step: 9, seconds: 0.118, TFLOP/s/device: 114.714, Tokens/s/device: 17291.018, 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