2026-04-15 22:00:34.077793: 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) I0415 22:00:34.194636 134853725224064 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 22:01:05.675190 134853725224064 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 22:01:05.675313 134853725224064 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 22:01:05.675349 134853725224064 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 I0415 22:01:05.675374 134853725224064 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 22:01:05.675398 134853725224064 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0415 22:01:05.675484 134853725224064 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 22:01:05.768467 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:06.181199 134853725224064 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 22:01:06.181543 134853725224064 checkpointing.py:677] Setting up checkpoint logger... I0415 22:01:06.181585 134853725224064 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 22:01:06.181624 134853725224064 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 22:01:06.182483 134853725224064 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 0x7aa57d7f8860>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 22:01:08.612393 134853725224064 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 22:01:08.612606 134853725224064 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 0x7a9f01538e90>}, handler_registry=None I0415 22:01:08.612909 134853725224064 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a9f01538e90>` 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`. I0415 22:01:08.612950 134853725224064 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9f01539a00>` 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`. I0415 22:01:08.612985 134853725224064 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 0x7a9f01538e90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a9f01538e90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9f01539a00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a9f01539a00>}). I0415 22:01:08.613326 134853725224064 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 22:01:08.613381 134853725224064 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7a9f0152d3a0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 22:01:08.740471 134853725224064 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints I0415 22:01:08.740740 134853725224064 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a9f01538230> I0415 22:01:08.740825 134853725224064 checkpointing.py:301] Checkpoint manager created! W0415 22:01:08.758086 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:08.948137 134853725224064 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 22:01:08.948316 134853725224064 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:01:09.040981 134853725224064 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 22:01:09.041080 134853725224064 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:01:09.055261 134853725224064 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 22:01:09.055313 134853725224064 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 22:01:09.075823 134853725224064 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:01:09.075880 134853725224064 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:01:09.090033 134853725224064 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:01:09.090100 134853725224064 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:01:09.104284 134853725224064 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 22:01:09.104337 134853725224064 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:01:09.126112 134853725224064 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 22:01:09.126173 134853725224064 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 22:01:09.144538 134853725224064 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 22:01:09.144593 134853725224064 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 22:01:09.330074 134853725224064 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 22:01:09.330177 134853725224064 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0415 22:01:09.382431 3278341 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 I0415 22:01:09.766482 134853725224064 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 22:01:09.766592 134853725224064 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 22:01:09.766632 134853725224064 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 22:01:09.766675 134853725224064 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 22:01:09.766713 134853725224064 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 22:01:09.766744 134853725224064 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 22:01:09.766782 134853725224064 maxtext_utils.py:1620] 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) I0415 22:01:09.766820 134853725224064 maxtext_utils.py:1620] 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') I0415 22:01:09.766848 134853725224064 maxtext_utils.py:1620] 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) I0415 22:01:09.766875 134853725224064 maxtext_utils.py:1620] 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) I0415 22:01:09.766910 134853725224064 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0415 22:01:09.766943 134853725224064 maxtext_utils.py:1620] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0415 22:01:10.191417 134853725224064 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 22:01:10.191509 134853725224064 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 22:01:10.204946 134853725224064 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 22:01:10.205002 134853725224064 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 22:01:10.850150 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:10.969599 134853725224064 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. I0415 22:01:10.970345 134853725224064 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 22:01:10.970529 134853725224064 metric_logger.py:289] number parameters: 1.104 billion W0415 22:01:11.837698 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:11.936745 134853725224064 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0415 22:01:12.054007 134853725224064 checkpointing.py:776] Waited 0.11723542213439941 seconds for step 0 to finish before starting checkpointing. I0415 22:01:12.054557 134853725224064 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 22:01:12.054714 134853725224064 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0415 22:01:12.055219 134853725224064 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0. I0415 22:01:12.155197 134853725224064 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 22:01:12.249336 134853725224064 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:01:12.249436 134853725224064 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 I0415 22:01:12.255995 134746248054336 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0 W0415 22:01:12.483352 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.497233 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.504258 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.515574 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.531013 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.541823 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.547154 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:12.551985 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:12.968821 134746237568576 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/items I0415 22:01:13.145611 134746206111296 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776290472878458258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776290472878458258, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0415 22:01:13.475076 3280468 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 22:01:13.475150 3280468 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 22:01:14.501885 134853725224064 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.253095s I0415 22:01:14.506436 134853725224064 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.252 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0415 22:01:14.506505 134853725224064 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.349797s (batch_requests_ready=0.088253s, total_serialization_initiated=2.257148s, others=0.004396s) I0415 22:01:14.506578 134853725224064 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.350813s (all_items=0.000031s, per_item={'items': '0.00003099'}, temp_paths=2.350782) I0415 22:01:14.508110 134746258540096 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:01:14.508213 134853725224064 async_checkpointer.py:561] Finished blocking save. Time taken: 2.453457s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0. I0415 22:01:14.508483 134853725224064 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:01:14.508656 134746122225216 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:01:14.508721 134853725224064 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776290472.0545425, 'wait_for_prev_duration_secs': 4.482269287109375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776290472.0547366, 'checkpointer_blocking_duration_secs': 2.4536688327789307, 'get_old_steps_start_time': 1776290474.508425, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1776290472.054452, 'checkpoint_manager_blocking_duration_secs': 2.4542434215545654} I0415 22:01:14.508853 134853725224064 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0415 22:01:14.508897 134853725224064 max_utils.py:750] Memstats: After params initialized: I0415 22:01:14.508948 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0415 22:01:14.508973 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 22:01:14.508992 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 22:01:14.509010 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 22:01:14.509027 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 22:01:14.509059 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 22:01:14.509079 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 22:01:14.509097 134853725224064 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 22:01:14.515310 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:14.521902 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 22:01:14.526784 3278341 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 22:01:14.746279 134853725224064 metric_logger.py:185] completed step: 0, seconds: 0.966, TFLOP/s/device: 14.071, Tokens/s/device: 2120.968, total_weights: 16384, loss: 10.874 I0415 22:01:14.747182 134853725224064 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/tensorboard/' I0415 22:01:14.862460 134853725224064 metric_logger.py:185] completed step: 1, seconds: 2.808, TFLOP/s/device: 4.839, Tokens/s/device: 729.393, total_weights: 16384, loss: 10.874 I0415 22:01:14.980978 134853725224064 metric_logger.py:185] completed step: 2, seconds: 0.005, TFLOP/s/device: 2470.836, Tokens/s/device: 372431.351, total_weights: 16384, loss: 10.276 I0415 22:01:15.099313 134853725224064 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.008, Tokens/s/device: 17486.040, total_weights: 16384, loss: 9.282 I0415 22:01:15.104356 134853725224064 checkpointing.py:772] Waiting for step 5 to finish before checkpoint... I0415 22:01:15.286000 134746164168256 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_0 I0415 22:01:15.335934 134853725224064 checkpointing.py:776] Waited 0.23152947425842285 seconds for step 5 to finish before starting checkpointing. I0415 22:01:15.336414 134853725224064 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:01:45.869219 134746153682496 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.361888s (commit=30.925199s, array_metadata_write=0.436689s) I0415 22:01:45.870433 134746258540096 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 374.827 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0415 22:01:45.870552 134746258540096 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.362230s. I0415 22:01:46.100530 134746258540096 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776290472878458258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0415 22:01:46.298859 134746258540096 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:01:46.479663 134746206111296 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776290472878458258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/_CHECKPOINT_METADATA I0415 22:01:46.739407 134746258540096 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 22:01:46.739611 134746258540096 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.597037s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/items I0415 22:01:46.740353 134746258540096 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0/items I0415 22:01:47.000977 134746258540096 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0 I0415 22:01:47.702793 134746258540096 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0`. I0415 22:01:47.703576 134746258540096 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.648828s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/0 I0415 22:01:47.703658 134746258540096 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.195338s. I0415 22:01:47.703777 134746122225216 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:01:47.703832 134746122225216 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:01:47.703898 134746122225216 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:01:47.703938 134746122225216 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0415 22:01:47.704132 134853725224064 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. W0415 22:01:47.704286 134853725224064 checkpoint_manager.py:1441] Waiting for previous save to complete took 32.367873 seconds. If this number is high, consider checkpointing less frequently. I0415 22:01:47.705114 134853725224064 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 5 I0415 22:01:47.705426 134853725224064 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5. I0415 22:01:47.900427 134853725224064 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:01:47.900532 134853725224064 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 I0415 22:01:47.904227 134746122225216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5 I0415 22:01:48.626904 134746132710976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5/items I0415 22:01:59.942080 134853725224064 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 12.042156s I0415 22:01:59.947432 134853725224064 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.017 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 12 seconds) (per-host) I0415 22:01:59.947497 134853725224064 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 12.135421s (batch_requests_ready=0.084952s, total_serialization_initiated=12.045331s, others=0.005138s) I0415 22:01:59.947607 134853725224064 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 12.136052s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=12.136033) I0415 22:01:59.949171 134746122225216 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:01:59.949304 134853725224064 async_checkpointer.py:561] Finished blocking save. Time taken: 12.244142s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5. I0415 22:01:59.949615 134853725224064 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:01:59.949796 134746258540096 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:01:59.949890 134853725224064 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776290475.3363876, 'wait_for_prev_duration_secs': 32.36787295341492, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776290507.705142, 'checkpointer_blocking_duration_secs': 12.244349479675293, 'get_old_steps_start_time': 1776290519.949533, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776290475.3363156, 'checkpoint_manager_blocking_duration_secs': 44.61354732513428} I0415 22:01:59.950071 134853725224064 checkpointing.py:408] Started an asynchronous checkpoint save for step 5 I0415 22:01:59.950730 134853725224064 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 114.743, Tokens/s/device: 17295.252, total_weights: 16384, loss: 8.327 I0415 22:01:59.959981 134853725224064 metric_logger.py:185] completed step: 5, seconds: 0.119, TFLOP/s/device: 114.535, Tokens/s/device: 17263.907, total_weights: 16384, loss: 7.367 I0415 22:02:00.075476 134853725224064 metric_logger.py:185] completed step: 6, seconds: 44.853, TFLOP/s/device: 0.303, Tokens/s/device: 45.660, total_weights: 16384, loss: 6.459 I0415 22:02:00.194250 134853725224064 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2177.425, Tokens/s/device: 328205.128, total_weights: 16384, loss: 5.622 I0415 22:02:00.312479 134853725224064 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.558, Tokens/s/device: 17568.843, total_weights: 16384, loss: 4.864 I0415 22:02:00.430456 134853725224064 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0415 22:02:00.431405 134853725224064 checkpointing.py:776] Waited 0.0009846687316894531 seconds for step 9 to finish before starting checkpointing. I0415 22:02:00.431674 134853725224064 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:02:00.786709 134746380174912 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_0 I0415 22:02:32.654376 134746132710976 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.706095s (commit=32.272191s, array_metadata_write=0.433904s) I0415 22:02:32.655759 134746122225216 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 281.797 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0415 22:02:32.655894 134746122225216 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.706545s. I0415 22:02:33.056575 134746122225216 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:02:33.510811 134746122225216 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 22:02:33.511020 134746122225216 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.594591s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5/items I0415 22:02:33.511822 134746122225216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5/items I0415 22:02:33.766370 134746122225216 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5 I0415 22:02:34.449139 134746122225216 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5`. I0415 22:02:34.449908 134746122225216 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.744752s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/5 I0415 22:02:34.449991 134746122225216 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.500644s. I0415 22:02:34.450147 134746258540096 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:02:34.450200 134746258540096 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:02:34.450261 134746258540096 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:02:34.450299 134746258540096 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0415 22:02:34.450478 134853725224064 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0415 22:02:34.450627 134853725224064 checkpoint_manager.py:1441] Waiting for previous save to complete took 34.018962 seconds. If this number is high, consider checkpointing less frequently. I0415 22:02:34.450877 134853725224064 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0415 22:02:34.451206 134853725224064 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9. I0415 22:02:34.625881 134746122225216 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9 I0415 22:02:34.651358 134853725224064 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 22:02:34.651464 134853725224064 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 I0415 22:02:35.373881 134746359203392 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9/items I0415 22:02:46.012344 134853725224064 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 11.361438s I0415 22:02:46.013000 134853725224064 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.076 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 11 seconds) (per-host) I0415 22:02:46.013074 134853725224064 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 11.472855s (batch_requests_ready=0.106929s, total_serialization_initiated=11.365404s, others=0.000522s) I0415 22:02:46.013154 134853725224064 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 11.473445s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=11.473431) I0415 22:02:46.019492 134746132710976 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 22:02:46.019644 134853725224064 async_checkpointer.py:561] Finished blocking save. Time taken: 11.568717s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9. I0415 22:02:46.019904 134853725224064 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0415 22:02:46.020088 134746258540096 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 22:02:46.020194 134853725224064 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776290520.4316452, 'wait_for_prev_duration_secs': 34.018961668014526, 'time_between_consecutive_saves_sec': 12.727691411972046, 'checkpointer_blocking_start_time': 1776290554.4509006, 'checkpointer_blocking_duration_secs': 11.568898677825928, 'get_old_steps_start_time': 1776290566.019815, 'get_old_steps_duration_secs': 3.6716461181640625e-05, 'checkpoint_manager_blocking_start_time': 1776290520.4316099, 'checkpoint_manager_blocking_duration_secs': 45.588550329208374} I0415 22:02:46.020342 134853725224064 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0415 22:02:46.020382 134853725224064 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 22:02:46.806944 134746369689152 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_0 I0415 22:03:22.721304 134746359203392 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.703037s (commit=36.239337s, array_metadata_write=0.463700s) I0415 22:03:22.722692 134746132710976 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 262.270 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 48 seconds) (per-host) I0415 22:03:22.722828 134746132710976 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.703121s. I0415 22:03:23.150784 134746132710976 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 22:03:23.584109 134746132710976 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 22:03:23.584326 134746132710976 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.587364s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9/items I0415 22:03:23.585135 134746132710976 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9/items I0415 22:03:23.832462 134746132710976 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9 I0415 22:03:24.523569 134746132710976 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9`. I0415 22:03:24.524363 134746132710976 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.073447s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_15_ocdbt_true/checkpoints/9 I0415 22:03:24.524438 134746132710976 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 38.504734s. I0415 22:03:24.524565 134746258540096 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 22:03:24.524619 134746258540096 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 22:03:24.524681 134746258540096 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0415 22:03:24.524725 134746258540096 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0415 22:03:24.525480 134853725224064 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. I0415 22:03:24.526614 134853725224064 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 22:03:24.528108 134853725224064 metric_logger.py:185] completed step: 9, seconds: 0.119, TFLOP/s/device: 114.341, Tokens/s/device: 17234.705, total_weights: 16384, loss: 4.189 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops