MaxView

2026-04-21 17:56:04.753686: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1776794164.763867 1681589 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1776794164.766968 1681589 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1776794164.776947 1681589 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776794164.776958 1681589 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776794164.776960 1681589 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776794164.776962 1681589 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
`rope_scaling`'s factor field must be a float >= 1, got 40
`rope_scaling`'s beta_fast field must be a float, got 32
`rope_scaling`'s beta_slow field must be a float, got 1
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'}
2026-04-21 17:56:08.436501: 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)
I0421 17:56:08.571310 124098523028608 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0421 17:56:38.189669 124098523028608 max_utils.py:800] System Information: Jax Version: 0.8.1
I0421 17:56:38.189836 124098523028608 max_utils.py:801] System Information: Jaxlib Version: 0.8.1
I0421 17:56:38.189870 124098523028608 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709
I0421 17:56:38.189901 124098523028608 train_utils.py:378] WARNING: 'dataset_path' might be pointing your local file system
I0421 17:56:38.189926 124098523028608 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0421 17:56:38.190057 124098523028608 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
I0421 17:56:38.820600 124098523028608 maxtext_utils.py:1732] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0421 17:56:38.921775 124098523028608 checkpointing.py:688] Setting up checkpoint logger...
I0421 17:56:38.921887 124098523028608 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0421 17:56:38.921966 124098523028608 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0421 17:56:38.922219 124098523028608 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 0x70dda6bae030>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0421 17:56:41.344489 124098523028608 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0421 17:56:41.344745 124098523028608 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 0x70dbfcd90fe0>}, handler_registry=None
I0421 17:56:41.345095 124098523028608 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70dbfcd90fe0>` 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`.
I0421 17:56:41.345143 124098523028608 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70dbfcda8d40>` 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`.
I0421 17:56:41.345175 124098523028608 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 0x70dbfcd90fe0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x70dbfcd90fe0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70dbfcda8d40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x70dbfcda8d40>}).
I0421 17:56:41.345468 124098523028608 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0421 17:56:41.345532 124098523028608 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x70dbfe1a8c20> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0421 17:56:41.474212 124098523028608 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints
I0421 17:56:41.474478 124098523028608 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/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x70dbfcda8470>
I0421 17:56:41.474570 124098523028608 checkpointing.py:302] Checkpoint manager created!
I0421 17:56:42.325492 124098523028608 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0421 17:56:42.325610 124098523028608 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
[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

I0421 17:56:43.186832 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown.
I0421 17:56:43.186927 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,).
I0421 17:56:43.197995 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown.
I0421 17:56:43.198060 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None).
I0421 17:56:43.219967 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown.
I0421 17:56:43.220023 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp').
I0421 17:56:43.231787 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown.
I0421 17:56:43.231912 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None).
I0421 17:56:43.254172 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown.
I0421 17:56:43.254230 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None).
I0421 17:56:43.265156 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown.
I0421 17:56:43.265205 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp').
I0421 17:56:43.297969 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown.
I0421 17:56:43.298035 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None).
I0421 17:56:43.309006 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown.
I0421 17:56:43.309068 124098523028608 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp').
I0421 17:56:43.480720 124098523028608 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0421 17:56:43.480818 124098523028608 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0421 17:56:43.484767 124098523028608 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0421 17:56:43.484820 124098523028608 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0421 17:56:43.497193 124098523028608 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 17:56:43.497247 124098523028608 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0421 17:56:43.509408 124098523028608 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 17:56:43.509459 124098523028608 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0421 17:56:43.527490 124098523028608 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 17:56:43.527557 124098523028608 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0421 17:56:43.539725 124098523028608 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 17:56:43.539780 124098523028608 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0421 17:56:43.552071 124098523028608 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 17:56:43.552123 124098523028608 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0421 17:56:43.572913 124098523028608 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0421 17:56:43.572973 124098523028608 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0421 17:56:43.588254 124098523028608 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0421 17:56:43.588309 124098523028608 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0421 17:56:52.202816 124098523028608 max_utils.py:791] Total memory size: 4.5 GB, Output size: 1.5 GB, Temp size: 2.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0421 17:56:52.206467 124098523028608 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0421 17:56:52.207929 124098523028608 metric_logger.py:301] number parameters: 1.104 billion
I0421 17:56:54.861941 124098523028608 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0421 17:56:55.289790 124098523028608 checkpointing.py:798] Waited 0.42783308029174805 seconds for step 0 to finish before starting checkpointing.
I0421 17:56:55.290293 124098523028608 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 17:56:55.290442 124098523028608 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0421 17:56:55.290848 124098523028608 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0.
I0421 17:56:55.372373 124098523028608 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0421 17:56:55.460297 124014297089600 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0
I0421 17:56:55.473060 124098523028608 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0421 17:56:55.473162 124098523028608 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
I0421 17:56:56.150734 124014286603840 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/items
I0421 17:56:56.296450 124014250952256 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776794216049246160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776794216049246160, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0421 17:56:56.667927 124098523028608 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.196883s
I0421 17:56:56.668395 124098523028608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 9.534 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host)
I0421 17:56:56.668447 124098523028608 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.294417s (batch_requests_ready=0.089108s, total_serialization_initiated=1.204968s, others=0.000341s)
I0421 17:56:56.668510 124098523028608 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.295572s (all_items=0.000021s, per_item={'items': '0.00002074'}, temp_paths=1.295551)
I0421 17:56:56.669337 124014141900352 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0421 17:56:56.669441 124098523028608 async_checkpointer.py:561] Finished blocking save. Time taken: 1.378960s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0.
I0421 17:56:56.669624 124098523028608 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 17:56:56.669806 124014131414592 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0421 17:56:56.669873 124098523028608 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776794215.290281, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776794215.2904632, 'checkpointer_blocking_duration_secs': 1.3790912628173828, 'get_old_steps_start_time': 1776794216.6695697, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776794215.290193, 'checkpoint_manager_blocking_duration_secs': 1.3796534538269043}
I0421 17:56:56.669999 124098523028608 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0421 17:56:56.670069 124098523028608 max_utils.py:750] 
Memstats: After params initialized:
I0421 17:56:56.670187 124098523028608 max_utils.py:756] 	Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0))
I0421 17:56:56.670215 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0421 17:56:56.670235 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0421 17:56:56.670254 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0421 17:56:56.670273 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0421 17:56:56.670290 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0421 17:56:56.670306 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0421 17:56:56.670322 124098523028608 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
I0421 17:56:56.706485 1683946 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0421 17:56:56.706579 1683946 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0421 17:56:57.116631 124098523028608 metric_logger.py:196] completed step: 0, seconds: 2.653, TFLOP/s/device: 20.488, Tokens/s/device: 3088.235, total_weights: 65536, loss: 10.880, lm_loss: 10.880, perplexity: 53086.324
I0421 17:56:57.117547 124098523028608 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/tensorboard/'
I0421 17:56:57.490869 124014297089600 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/items/array_metadatas/process_0
I0421 17:56:57.529912 124098523028608 metric_logger.py:196] completed step: 1, seconds: 2.250, TFLOP/s/device: 24.153, Tokens/s/device: 3640.630, total_weights: 65536, loss: 10.880, lm_loss: 10.880, perplexity: 53086.324
I0421 17:56:57.952605 124098523028608 metric_logger.py:196] completed step: 2, seconds: 0.019, TFLOP/s/device: 2788.963, Tokens/s/device: 420382.819, total_weights: 65536, loss: 10.270, lm_loss: 10.270, perplexity: 28842.992
I0421 17:56:58.369409 124098523028608 metric_logger.py:196] completed step: 3, seconds: 0.406, TFLOP/s/device: 133.714, Tokens/s/device: 20154.901, total_weights: 65536, loss: 9.741, lm_loss: 9.741, perplexity: 16998.510
I0421 17:57:25.666921 124098523028608 metric_logger.py:196] completed step: 4, seconds: 0.432, TFLOP/s/device: 125.910, Tokens/s/device: 18978.515, total_weights: 65536, loss: 9.284, lm_loss: 9.284, perplexity: 10759.190
I0421 17:57:25.682484 124098523028608 metric_logger.py:196] completed step: 5, seconds: 0.424, TFLOP/s/device: 128.285, Tokens/s/device: 19336.534, total_weights: 65536, loss: 8.897, lm_loss: 8.897, perplexity: 7313.255
I0421 17:57:26.112931 124098523028608 metric_logger.py:196] completed step: 6, seconds: 27.286, TFLOP/s/device: 1.992, Tokens/s/device: 300.229, total_weights: 65536, loss: 8.598, lm_loss: 8.598, perplexity: 5421.259
I0421 17:57:26.540571 124098523028608 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 6170.358, Tokens/s/device: 930063.579, total_weights: 65536, loss: 8.390, lm_loss: 8.390, perplexity: 4402.202
I0421 17:57:26.970514 124098523028608 metric_logger.py:196] completed step: 8, seconds: 0.431, TFLOP/s/device: 126.027, Tokens/s/device: 18996.206, total_weights: 65536, loss: 8.261, lm_loss: 8.261, perplexity: 3868.122
I0421 17:57:27.397379 124098523028608 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0421 17:57:27.398514 124098523028608 checkpointing.py:798] Waited 0.0011458396911621094 seconds for step 9 to finish before starting checkpointing.
I0421 17:57:27.398828 124098523028608 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 17:57:36.710416 124014152386112 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.000293s (commit=39.559222s, array_metadata_write=0.441071s)
I0421 17:57:36.711810 124014141900352 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 305.697 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host)
I0421 17:57:36.711948 124014141900352 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.042455s.
I0421 17:57:36.950688 124014141900352 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776794216049246160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0421 17:57:37.136700 124014141900352 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0421 17:57:37.356753 124014250952256 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776794216049246160, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0421 17:57:37.565202 124014141900352 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0421 17:57:37.565924 124014141900352 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.578054s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/items
I0421 17:57:37.566703 124014141900352 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0/items
I0421 17:57:37.821202 124014141900352 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0
I0421 17:57:38.512007 124014141900352 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0`.
I0421 17:57:38.512770 124014141900352 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.222291s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/0
I0421 17:57:38.512854 124014141900352 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.843367s.
I0421 17:57:38.512991 124014131414592 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 17:57:38.513041 124014131414592 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0421 17:57:38.513119 124014131414592 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0421 17:57:38.513170 124014131414592 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0421 17:57:38.513521 124098523028608 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.
W0421 17:57:38.513899 124098523028608 checkpoint_manager.py:1441] Waiting for previous save to complete took 11.115106 seconds. If this number is high, consider checkpointing less frequently.
I0421 17:57:38.515117 124098523028608 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0421 17:57:38.515445 124098523028608 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9.
I0421 17:57:38.705774 124014131414592 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9
I0421 17:57:38.705960 124098523028608 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0421 17:57:38.706234 124098523028608 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
I0421 17:57:39.459779 124014078985792 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9/items
I0421 17:57:41.291573 124098523028608 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.587349s
I0421 17:57:41.297824 124098523028608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.583 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host)
I0421 17:57:41.297894 124098523028608 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.692527s (batch_requests_ready=0.090549s, total_serialization_initiated=2.595845s, others=0.006133s)
I0421 17:57:41.297976 124098523028608 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.693139s (all_items=0.000024s, per_item={'items': '0.00002408'}, temp_paths=2.693115)
I0421 17:57:41.299696 124014131414592 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0421 17:57:41.299784 124098523028608 async_checkpointer.py:561] Finished blocking save. Time taken: 2.784619s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9.
I0421 17:57:41.300016 124098523028608 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 17:57:41.300271 124014141900352 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0421 17:57:41.300391 124098523028608 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776794247.3987613, 'wait_for_prev_duration_secs': 11.115106344223022, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776794258.5151427, 'checkpointer_blocking_duration_secs': 2.78477144241333, 'get_old_steps_start_time': 1776794261.299931, 'get_old_steps_duration_secs': 4.363059997558594e-05, 'checkpoint_manager_blocking_start_time': 1776794247.3986986, 'checkpoint_manager_blocking_duration_secs': 13.901665687561035}
I0421 17:57:41.300516 124098523028608 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0421 17:57:41.300550 124098523028608 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 17:57:42.073923 124014297089600 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9/items/array_metadatas/process_0
I0421 17:58:22.788841 124014078985792 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.490147s (commit=41.035607s, array_metadata_write=0.454540s)
I0421 17:58:22.790130 124014131414592 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 285.999 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host)
I0421 17:58:22.790194 124014131414592 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.490348s.
I0421 17:58:23.279346 124014131414592 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0421 17:58:23.704908 124014131414592 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0421 17:58:23.705728 124014131414592 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.576218s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9/items
I0421 17:58:23.706459 124014131414592 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9/items
I0421 17:58:23.951252 124014131414592 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9
I0421 17:58:24.625718 124014131414592 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9`.
I0421 17:58:24.626478 124014131414592 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.111319s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_trainstate_and_training_loop_20260421_161934/nnx_feat_nnx_trainstate_and_training_loop_20260421_161934_06_grad_accum/checkpoints/9
I0421 17:58:24.626554 124014131414592 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.326710s.
I0421 17:58:24.626777 124014141900352 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 17:58:24.626906 124014141900352 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0421 17:58:24.626979 124014141900352 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0421 17:58:24.627023 124014141900352 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0421 17:58:24.627222 124098523028608 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.
I0421 17:58:24.627431 124098523028608 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 17:58:24.628891 124098523028608 metric_logger.py:196] completed step: 9, seconds: 0.428, TFLOP/s/device: 126.899, Tokens/s/device: 19127.584, total_weights: 65536, loss: 8.185, lm_loss: 8.185, perplexity: 3585.401
Per train step:
 Total TFLOPs: 54.35 
 split as 93.93% learnable weight flops and 6.07% attention flops