2026-04-16 13:14:42.246280: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)
I0416 13:14:42.364295 136779219651712 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 13:15:37.229948 136779219651712 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 13:15:37.230067 136779219651712 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 13:15:37.230102 136779219651712 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894
I0416 13:15:37.230127 136779219651712 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0416 13:15:37.230148 136779219651712 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 13:15:37.230231 136779219651712 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 13:15:37.321458 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:37.705374 136779219651712 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 13:15:37.705688 136779219651712 checkpointing.py:677] Setting up checkpoint logger...
I0416 13:15:37.705734 136779219651712 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 13:15:37.705781 136779219651712 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 13:15:37.706315 136779219651712 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 0x7c65ce1f4b00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 13:15:40.158115 136779219651712 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0416 13:15:40.158285 136779219651712 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 0x7c5f51f39ca0>}, handler_registry=None
I0416 13:15:40.158540 136779219651712 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c5f51f39ca0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`.
I0416 13:15:40.158579 136779219651712 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5f51f3a330>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`.
I0416 13:15:40.158616 136779219651712 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 0x7c5f51f39ca0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c5f51f39ca0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5f51f3a330>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c5f51f3a330>}).
I0416 13:15:40.158904 136779219651712 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 13:15:40.158956 136779219651712 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7c5f51f2d120> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 13:15:40.284085 136779219651712 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints
I0416 13:15:40.284298 136779219651712 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_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c5f51f3a1b0>
I0416 13:15:40.284377 136779219651712 checkpointing.py:301] Checkpoint manager created!
W0416 13:15:40.300506 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:40.486228 136779219651712 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 13:15:40.486426 136779219651712 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:15:40.582884 136779219651712 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 13:15:40.582957 136779219651712 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:15:40.597121 136779219651712 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 13:15:40.597168 136779219651712 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 13:15:40.617619 136779219651712 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 13:15:40.617673 136779219651712 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:15:40.631775 136779219651712 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 13:15:40.631823 136779219651712 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:15:40.645914 136779219651712 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 13:15:40.645961 136779219651712 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:15:40.667512 136779219651712 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 13:15:40.667568 136779219651712 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 13:15:40.685791 136779219651712 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 13:15:40.685844 136779219651712 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 13:15:40.868524 136779219651712 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 13:15:40.868633 136779219651712 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
W0416 13:15:40.920538 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
[DECOUPLED NO-OP] gcs_storage: using stubs.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] workload_monitor: using stub.
[DECOUPLED NO-OP] vertex_tensorboard: using stub.
fsdp: 8
I0416 13:15:41.788830 136779219651712 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 13:15:41.788950 136779219651712 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)
I0416 13:15:41.788991 136779219651712 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)
I0416 13:15:41.789035 136779219651712 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')
I0416 13:15:41.789092 136779219651712 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)
I0416 13:15:41.789125 136779219651712 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)
I0416 13:15:41.789165 136779219651712 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)
I0416 13:15:41.789203 136779219651712 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')
I0416 13:15:41.789231 136779219651712 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)
I0416 13:15:41.789256 136779219651712 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)
I0416 13:15:41.789288 136779219651712 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 13:15:41.789321 136779219651712 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 13:15:42.641393 136779219651712 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 13:15:42.641473 136779219651712 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 13:15:42.654908 136779219651712 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 13:15:42.654955 136779219651712 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 13:15:43.412387 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:43.555142 136779219651712 max_utils.py:791] Total memory size: 2.4 GB, Output size: 0.5 GB, Temp size: 1.9 GB, Argument size: 0.5 GB, Host temp size: 0.0 GB.
I0416 13:15:44.802643 136779219651712 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 13:15:44.802874 136779219651712 metric_logger.py:289] number parameters: 1.104 billion
W0416 13:15:45.603149 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:45.722770 136779219651712 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0416 13:15:45.911529 136779219651712 checkpointing.py:776] Waited 0.1887524127960205 seconds for step 0 to finish before starting checkpointing.
I0416 13:15:45.911956 136779219651712 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 13:15:45.912133 136779219651712 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 13:15:45.912484 136779219651712 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0.
I0416 13:15:46.015701 136779219651712 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 13:15:46.107911 136779219651712 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 13:15:46.108059 136671945295424 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0
I0416 13:15:46.108451 136779219651712 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
W0416 13:15:46.323154 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.335887 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.349561 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.367064 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.373603 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.378665 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.384173 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.388678 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:46.560935 136779219651712 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.453717s
I0416 13:15:46.561520 136779219651712 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 22.655 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 544 milliseconds) (per-host)
I0416 13:15:46.561675 136779219651712 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.545026s (batch_requests_ready=0.087070s, total_serialization_initiated=0.457472s, others=0.000485s)
I0416 13:15:46.561800 136779219651712 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.545629s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.545610)
I0416 13:15:46.562582 136671882380864 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 13:15:46.562676 136779219651712 async_checkpointer.py:561] Finished blocking save. Time taken: 0.650500s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0.
I0416 13:15:46.562862 136779219651712 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 13:15:46.563000 136671955781184 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 13:15:46.563106 136779219651712 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776345345.911943, 'wait_for_prev_duration_secs': 4.029273986816406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776345345.9121556, 'checkpointer_blocking_duration_secs': 0.6506187915802002, 'get_old_steps_start_time': 1776345346.5627978, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1776345345.911863, 'checkpoint_manager_blocking_duration_secs': 0.6512105464935303}
I0416 13:15:46.563189 136779219651712 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0416 13:15:46.563230 136779219651712 max_utils.py:750]
Memstats: After params initialized:
I0416 13:15:46.563332 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_0(process=0,(0,0,0,0))
I0416 13:15:46.563362 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_1(process=0,(1,0,0,0))
I0416 13:15:46.563386 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_2(process=0,(0,1,0,0))
I0416 13:15:46.563406 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_3(process=0,(1,1,0,0))
I0416 13:15:46.563426 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_4(process=0,(0,2,0,0))
I0416 13:15:46.563444 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_5(process=0,(1,2,0,0))
I0416 13:15:46.563465 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_6(process=0,(0,3,0,0))
I0416 13:15:46.563489 136779219651712 max_utils.py:756] Using (GB) 0.56 / 31.25 (1.792000%) on TPU_7(process=0,(1,3,0,0))
W0416 13:15:46.569393 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.575414 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 13:15:46.580178 3859595 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 13:15:46.923849 136779219651712 metric_logger.py:185] completed step: 0, seconds: 0.919, TFLOP/s/device: 14.778, Tokens/s/device: 2227.557, total_weights: 16384, loss: 10.874
I0416 13:15:46.924489 136779219651712 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/tensorboard/'
I0416 13:15:47.106500 136779219651712 metric_logger.py:185] completed step: 1, seconds: 1.199, TFLOP/s/device: 11.334, Tokens/s/device: 1708.329, total_weights: 16384, loss: 10.874
I0416 13:15:47.123812 136671934809664 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/items
I0416 13:15:47.295701 136779219651712 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 2075.004, Tokens/s/device: 312767.257, total_weights: 16384, loss: 9.775
I0416 13:15:47.298125 136671792203328 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776345347021004053, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776345347021004053, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 13:15:47.484622 136779219651712 metric_logger.py:185] completed step: 3, seconds: 0.185, TFLOP/s/device: 73.615, Tokens/s/device: 11096.121, total_weights: 16384, loss: 8.838
I0416 13:15:47.608288 3862414 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 13:15:47.608361 3862414 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 13:15:48.611935 136671903352384 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_0
I0416 13:16:07.543538 136779219651712 metric_logger.py:185] completed step: 4, seconds: 0.189, TFLOP/s/device: 71.867, Tokens/s/device: 10832.597, total_weights: 16384, loss: 7.965
I0416 13:16:07.551074 136779219651712 metric_logger.py:185] completed step: 5, seconds: 0.189, TFLOP/s/device: 72.054, Tokens/s/device: 10860.746, total_weights: 16384, loss: 7.233
I0416 13:16:07.734317 136779219651712 metric_logger.py:185] completed step: 6, seconds: 20.060, TFLOP/s/device: 0.677, Tokens/s/device: 102.096, total_weights: 16384, loss: 6.677
I0416 13:16:07.923512 136779219651712 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 2958.225, Tokens/s/device: 445895.929, total_weights: 16384, loss: 6.293
I0416 13:16:08.112730 136779219651712 metric_logger.py:185] completed step: 8, seconds: 0.184, TFLOP/s/device: 73.786, Tokens/s/device: 11121.852, total_weights: 16384, loss: 6.055
I0416 13:16:08.301658 136779219651712 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0416 13:16:08.306775 136779219651712 checkpointing.py:776] Waited 0.005154848098754883 seconds for step 9 to finish before starting checkpointing.
I0416 13:16:08.307061 136779219651712 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 13:16:11.809988 136671892866624 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.206827s (commit=23.771369s, array_metadata_write=0.435458s)
I0416 13:16:11.811028 136671882380864 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 489.906 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host)
I0416 13:16:11.811136 136671882380864 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.248417s.
I0416 13:16:12.055613 136671882380864 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776345347021004053, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 13:16:12.253705 136671882380864 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0416 13:16:12.444271 136671792203328 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776345347021004053, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 13:16:12.693214 136671882380864 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 13:16:12.693392 136671882380864 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.595587s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/items
I0416 13:16:12.694082 136671882380864 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0/items
I0416 13:16:12.931856 136671882380864 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0
I0416 13:16:13.614238 136671882380864 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0`.
I0416 13:16:13.614860 136671882380864 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 27.702691s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/0
I0416 13:16:13.614933 136671882380864 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 27.052215s.
I0416 13:16:13.615072 136671955781184 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 13:16:13.615121 136671955781184 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 13:16:13.615172 136671955781184 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 13:16:13.615212 136671955781184 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 13:16:13.615350 136779219651712 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0416 13:16:13.615477 136779219651712 checkpoint_manager.py:1441] Waiting for previous save to complete took 5.308441 seconds. If this number is high, consider checkpointing less frequently.
I0416 13:16:13.616311 136779219651712 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 13:16:13.616595 136779219651712 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9.
I0416 13:16:13.771138 136671955781184 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9
I0416 13:16:13.789467 136779219651712 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 13:16:13.789560 136779219651712 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0416 13:16:14.052934 136779219651712 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.263880s
I0416 13:16:14.053279 136779219651712 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 34.372 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 359 milliseconds) (per-host)
I0416 13:16:14.053332 136779219651712 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.359120s (batch_requests_ready=0.092016s, total_serialization_initiated=0.266878s, others=0.000227s)
I0416 13:16:14.053396 136779219651712 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.359632s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.359615)
I0416 13:16:14.054005 136671829952064 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 13:16:14.054108 136779219651712 async_checkpointer.py:561] Finished blocking save. Time taken: 0.437751s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9.
I0416 13:16:14.054286 136779219651712 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 13:16:14.054437 136671819466304 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 13:16:14.054533 136779219651712 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776345368.307008, 'wait_for_prev_duration_secs': 5.308441400527954, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776345373.6163368, 'checkpointer_blocking_duration_secs': 0.4378647804260254, 'get_old_steps_start_time': 1776345374.0542183, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776345368.3069668, 'checkpoint_manager_blocking_duration_secs': 5.747531175613403}
I0416 13:16:14.054625 136779219651712 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0416 13:16:14.054658 136779219651712 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 13:16:14.522845 136671892866624 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9/items
I0416 13:16:15.747685 136671850923584 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_0
I0416 13:16:47.800601 136671840437824 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 32.819205s (commit=32.370612s, array_metadata_write=0.448592s)
I0416 13:16:47.801661 136671829952064 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 370.499 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 34 seconds) (per-host)
I0416 13:16:47.801720 136671829952064 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.747568s.
I0416 13:16:48.202087 136671829952064 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0416 13:16:48.639631 136671829952064 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 13:16:48.639812 136671829952064 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.575125s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9/items
I0416 13:16:48.640485 136671829952064 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9/items
I0416 13:16:48.872135 136671829952064 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9
I0416 13:16:49.548208 136671829952064 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9`.
I0416 13:16:49.548851 136671829952064 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.932499s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_11_optimizer_offload_true/checkpoints/9
I0416 13:16:49.548929 136671829952064 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.494779s.
I0416 13:16:49.549077 136671819466304 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 13:16:49.549130 136671819466304 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 13:16:49.549187 136671819466304 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 13:16:49.549226 136671819466304 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 13:16:49.549467 136779219651712 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0416 13:16:49.549953 136779219651712 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 13:16:49.550714 136779219651712 metric_logger.py:185] completed step: 9, seconds: 0.189, TFLOP/s/device: 71.820, Tokens/s/device: 10825.497, total_weights: 16384, loss: 5.915
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops