2026-04-16 03:40:52.858939: 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 03:40:52.977800 124597500406912 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 03:41:45.174058 124597500406912 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 03:41:45.174169 124597500406912 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 03:41:45.174205 124597500406912 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 03:41:45.174230 124597500406912 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 03:41:45.174325 124597500406912 train.py:818] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 03:41:45.268335 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:41:45.673098 124597500406912 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 03:41:45.673269 124597500406912 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 03:41:45.673492 124597500406912 checkpointing.py:688] Setting up checkpoint logger...
I0416 03:41:45.673530 124597500406912 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 03:41:45.673573 124597500406912 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 03:41:45.674127 124597500406912 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 0x715186df4fb0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 03:41:48.020818 124597500406912 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 03:41:48.021224 124597500406912 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 0x714b0b3cea80>}, handler_registry=None
I0416 03:41:48.021769 124597500406912 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x714b0b3cea80>` 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 03:41:48.021815 124597500406912 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714b0ab3b110>` 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 03:41:48.021844 124597500406912 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 0x714b0b3cea80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x714b0b3cea80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714b0ab3b110>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714b0ab3b110>}).
I0416 03:41:48.022611 124597500406912 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 03:41:48.022697 124597500406912 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x714b0ab460c0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 03:41:48.142648 124597500406912 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints
I0416 03:41:48.142910 124597500406912 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_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x714b0af14a10>
I0416 03:41:48.143005 124597500406912 checkpointing.py:302] Checkpoint manager created!
I0416 03:41:48.227772 124597500406912 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0416 03:41:48.230905 124597500406912 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0.
I0416 03:41:48.285695 124597500406912 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0416 03:41:48.318863 124597500406912 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 03:41:48.318932 124597500406912 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 03:41:49.322125 124597500406912 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 03:41:49.322216 124597500406912 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0416 03:41:49.423796 124597500406912 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 03:41:49.423875 124597500406912 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0416 03:41:49.438748 124597500406912 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 03:41:49.438797 124597500406912 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0416 03:41:49.461073 124597500406912 attentions.py:1154] attentions/query Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 03:41:49.461128 124597500406912 attentions.py:1154] attentions/query Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0416 03:41:49.476083 124597500406912 attentions.py:1155] attentions/key Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 03:41:49.476130 124597500406912 attentions.py:1155] attentions/key Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0416 03:41:49.491013 124597500406912 attentions.py:1156] attentions/value Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 03:41:49.491076 124597500406912 attentions.py:1156] attentions/value Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0416 03:41:49.513108 124597500406912 attentions.py:1197] attentions/out Logical: bfloat16[8,128,16,128]...................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 03:41:49.513166 124597500406912 attentions.py:1197] attentions/out Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0416 03:41:49.536426 124597500406912 linears.py:525] linears/x Logical: bfloat16[8,128,7168]........................................ ('activation_batch', 'activation_length', 'activation_mlp').
I0416 03:41:49.536479 124597500406912 linears.py:525] linears/x Physical: bfloat16[8,128,7168]........................................ ('fsdp', None, None).
I0416 03:41:49.736935 124597500406912 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 03:41:49.737028 124597500406912 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0416 03:41:49.787441 3629703 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 03:41:50.189612 124597500406912 maxtext_utils.py:1790] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 03:41:50.189727 124597500406912 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 03:41:50.189769 124597500406912 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 03:41:50.189815 124597500406912 maxtext_utils.py:1790] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0416 03:41:50.189856 124597500406912 maxtext_utils.py:1790] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 03:41:50.189884 124597500406912 maxtext_utils.py:1790] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 03:41:50.189923 124597500406912 maxtext_utils.py:1790] 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 03:41:50.189961 124597500406912 maxtext_utils.py:1790] 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 03:41:50.189989 124597500406912 maxtext_utils.py:1790] 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 03:41:50.190016 124597500406912 maxtext_utils.py:1790] 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 03:41:50.190123 124597500406912 maxtext_utils.py:1790] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 03:41:50.190167 124597500406912 maxtext_utils.py:1790] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 03:41:50.649940 124597500406912 train.py:157] train/xent Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 03:41:50.650021 124597500406912 train.py:157] train/xent Physical: float32[8,128].............................................. ('fsdp', None).
I0416 03:41:50.664626 124597500406912 train.py:164] train/z_loss Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 03:41:50.664678 124597500406912 train.py:164] train/z_loss Physical: float32[8,128].............................................. ('fsdp', None).
W0416 03:41:50.924707 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:41:51.015673 124597500406912 max_utils.py:791] Total memory size: 2.5 GB, Output size: 1.5 GB, Temp size: 1.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0416 03:41:51.016149 124597500406912 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 03:41:51.016293 124597500406912 metric_logger.py:289] number parameters: 1.104 billion
W0416 03:41:54.096142 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:41:54.199623 124597500406912 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0416 03:41:54.265468 124597500406912 checkpointing.py:798] Waited 0.0658423900604248 seconds for step 0 to finish before starting checkpointing.
I0416 03:41:54.266137 124597500406912 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 03:41:54.266334 124597500406912 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 03:41:54.266752 124597500406912 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0.
I0416 03:41:54.342280 124597500406912 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 03:41:54.440079 124597500406912 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 03:41:54.440220 124484189161024 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0
I0416 03:41:54.440607 124597500406912 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 03:41:54.657842 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.668376 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.685165 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.698647 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.706791 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.712468 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.717615 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.723481 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:41:54.914586 124597500406912 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.475259s
I0416 03:41:54.914956 124597500406912 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 21.589 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 571 milliseconds) (per-host)
I0416 03:41:54.915007 124597500406912 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.571730s (batch_requests_ready=0.092470s, total_serialization_initiated=0.478995s, others=0.000265s)
I0416 03:41:54.915079 124597500406912 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.572327s (all_items=0.000040s, per_item={'items': '0.00003982'}, temp_paths=0.572287)
I0416 03:41:54.915834 124484117857856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 03:41:54.915927 124597500406912 async_checkpointer.py:561] Finished blocking save. Time taken: 0.649549s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0.
I0416 03:41:54.916174 124597500406912 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 03:41:54.916324 124484199646784 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 03:41:54.916406 124597500406912 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776310914.2661164, 'wait_for_prev_duration_secs': 5.030632019042969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776310914.266356, 'checkpointer_blocking_duration_secs': 0.6496706008911133, 'get_old_steps_start_time': 1776310914.9161038, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1776310914.265924, 'checkpoint_manager_blocking_duration_secs': 0.650449275970459}
I0416 03:41:54.916487 124597500406912 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0416 03:41:54.916539 124597500406912 max_utils.py:750]
Memstats: After params initialized:
I0416 03:41:54.916618 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_0(process=0,(0,0,0,0))
I0416 03:41:54.916644 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_1(process=0,(1,0,0,0))
I0416 03:41:54.916671 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_2(process=0,(0,1,0,0))
I0416 03:41:54.916688 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_3(process=0,(1,1,0,0))
I0416 03:41:54.916704 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_4(process=0,(0,2,0,0))
I0416 03:41:54.916720 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_5(process=0,(1,2,0,0))
I0416 03:41:54.916738 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_6(process=0,(0,3,0,0))
I0416 03:41:54.916754 124597500406912 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_7(process=0,(1,3,0,0))
W0416 03:41:54.922552 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.928200 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 03:41:54.932812 3629703 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 03:41:55.236238 124597500406912 metric_logger.py:185] completed step: 0, seconds: 3.183, TFLOP/s/device: 0.252, Tokens/s/device: 40.216, total_weights: 963, loss: 10.834
I0416 03:41:55.237107 124597500406912 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/tensorboard/'
I0416 03:41:55.297742 124597500406912 metric_logger.py:185] completed step: 1, seconds: 1.035, TFLOP/s/device: 0.774, Tokens/s/device: 123.696, total_weights: 945, loss: 10.747
I0416 03:41:55.327146 124484178675264 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/items
I0416 03:41:55.361351 124597500406912 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 81.689, Tokens/s/device: 13055.896, total_weights: 962, loss: 10.246
I0416 03:41:55.425110 124597500406912 metric_logger.py:185] completed step: 3, seconds: 0.062, TFLOP/s/device: 12.904, Tokens/s/device: 2062.387, total_weights: 975, loss: 9.829
I0416 03:41:55.464454 124484027680320 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310915223152379, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776310915223152379, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:41:55.977982 3632432 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 03:41:55.978056 3632432 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 03:41:56.715623 124484138829376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/items/array_metadatas/process_0
I0416 03:42:05.951407 124597500406912 metric_logger.py:185] completed step: 4, seconds: 0.064, TFLOP/s/device: 12.548, Tokens/s/device: 2005.547, total_weights: 962, loss: 9.393
I0416 03:42:05.960347 124597500406912 metric_logger.py:185] completed step: 5, seconds: 0.064, TFLOP/s/device: 12.488, Tokens/s/device: 1995.821, total_weights: 958, loss: 9.248
I0416 03:42:06.022659 124597500406912 metric_logger.py:185] completed step: 6, seconds: 10.525, TFLOP/s/device: 0.076, Tokens/s/device: 12.161, total_weights: 930, loss: 9.424
I0416 03:42:06.086421 124597500406912 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 132.267, Tokens/s/device: 21139.554, total_weights: 935, loss: 9.317
I0416 03:42:06.150028 124597500406912 metric_logger.py:185] completed step: 8, seconds: 0.063, TFLOP/s/device: 12.689, Tokens/s/device: 2028.012, total_weights: 930, loss: 9.478
I0416 03:42:06.213391 124597500406912 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0416 03:42:06.214154 124597500406912 checkpointing.py:798] Waited 0.0007812976837158203 seconds for step 9 to finish before starting checkpointing.
I0416 03:42:06.214425 124597500406912 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 03:42:20.064373 124484128343616 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.089268s (commit=23.675793s, array_metadata_write=0.413476s)
I0416 03:42:20.065360 124484117857856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 491.283 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 25 seconds) (per-host)
I0416 03:42:20.065473 124484117857856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.149501s.
I0416 03:42:20.313378 124484117857856 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310915223152379, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:42:20.511117 124484117857856 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 03:42:20.710038 124484027680320 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776310915223152379, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0416 03:42:20.915481 124484117857856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 03:42:20.915664 124484117857856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.558500s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/items
I0416 03:42:20.916368 124484117857856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0/items
I0416 03:42:21.156512 124484117857856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0
I0416 03:42:21.842072 124484117857856 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0`.
I0416 03:42:21.842703 124484117857856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 27.576331s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/0
I0416 03:42:21.842783 124484117857856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 26.926813s.
I0416 03:42:21.842880 124484199646784 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 03:42:21.842931 124484199646784 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 03:42:21.842988 124484199646784 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 03:42:21.843034 124484199646784 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 03:42:21.843183 124597500406912 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 03:42:21.843311 124597500406912 checkpoint_manager.py:1441] Waiting for previous save to complete took 15.628898 seconds. If this number is high, consider checkpointing less frequently.
I0416 03:42:21.844032 124597500406912 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 03:42:21.844355 124597500406912 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9.
I0416 03:42:22.017893 124484199646784 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9
I0416 03:42:22.027899 124597500406912 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 03:42:22.027992 124597500406912 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 03:42:22.704167 124484178675264 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9/items
I0416 03:42:35.300006 124597500406912 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 13.272514s
I0416 03:42:35.304803 124597500406912 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 945.266 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 13 seconds) (per-host)
I0416 03:42:35.304867 124597500406912 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 13.368583s (batch_requests_ready=0.088397s, total_serialization_initiated=13.275530s, others=0.004657s)
I0416 03:42:35.304944 124597500406912 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 13.369099s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=13.369081)
I0416 03:42:35.306376 124484199646784 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 03:42:35.306458 124597500406912 async_checkpointer.py:561] Finished blocking save. Time taken: 13.462360s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9.
I0416 03:42:35.306655 124597500406912 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 03:42:35.306816 124484117857856 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 03:42:35.306918 124597500406912 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776310926.2143848, 'wait_for_prev_duration_secs': 15.62889814376831, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776310941.8440797, 'checkpointer_blocking_duration_secs': 13.46248483657837, 'get_old_steps_start_time': 1776310955.306584, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776310926.2143483, 'checkpoint_manager_blocking_duration_secs': 29.092540502548218}
I0416 03:42:35.307040 124597500406912 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0416 03:42:35.307090 124597500406912 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 03:42:36.048927 124484138829376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9/items/array_metadatas/process_0
I0416 03:43:10.936676 124484178675264 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.631171s (commit=35.196445s, array_metadata_write=0.434725s)
I0416 03:43:10.937689 124484199646784 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 257.887 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 49 seconds) (per-host)
I0416 03:43:10.937832 124484199646784 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.631323s.
I0416 03:43:11.364537 124484199646784 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 03:43:11.772773 124484199646784 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 03:43:11.773468 124484199646784 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.554171s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9/items
I0416 03:43:11.774123 124484199646784 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9/items
I0416 03:43:12.011213 124484199646784 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9
I0416 03:43:12.668530 124484199646784 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9`.
I0416 03:43:12.669170 124484199646784 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.825078s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260416_033532/linen_feat_nnx_trainstate_and_training_loop_20260416_033532_03_dropout/checkpoints/9
I0416 03:43:12.669246 124484199646784 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.362745s.
I0416 03:43:12.669411 124484117857856 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 03:43:12.669512 124484117857856 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 03:43:12.669572 124484117857856 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 03:43:12.669642 124484117857856 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 03:43:12.669850 124597500406912 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 03:43:12.669970 124597500406912 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 03:43:12.670547 124597500406912 metric_logger.py:185] completed step: 9, seconds: 0.064, TFLOP/s/device: 12.604, Tokens/s/device: 2014.352, total_weights: 948, loss: 9.141
Per train step:
Total TFLOPs: 0.80
split as 99.60% learnable weight flops and 0.40% attention flops