2026-04-20 14:28:57.712938: 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)
I0420 14:28:57.830459 133857053895808 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0420 14:29:50.880501 133857053895808 max_utils.py:800] System Information: Jax Version: 0.9.2
I0420 14:29:50.880620 133857053895808 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0420 14:29:50.880656 133857053895808 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
I0420 14:29:50.880683 133857053895808 train_utils.py:365] WARNING: 'dataset_path' might be pointing your local file system
I0420 14:29:50.880774 133857053895808 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
I0420 14:29:51.397536 133857053895808 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0420 14:29:51.397723 133857053895808 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0420 14:29:51.397911 133857053895808 checkpointing.py:688] Setting up checkpoint logger...
I0420 14:29:51.397950 133857053895808 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0420 14:29:51.397991 133857053895808 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0420 14:29:51.398480 133857053895808 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 0x79bd6f3f8e90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0420 14:29:53.809976 133857053895808 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0420 14:29:53.810220 133857053895808 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 0x79b6f015c0b0>}, handler_registry=None
I0420 14:29:53.810484 133857053895808 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b6f015c0b0>` 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`.
I0420 14:29:53.810527 133857053895808 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b6f015c4d0>` 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`.
I0420 14:29:53.810557 133857053895808 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 0x79b6f015c0b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b6f015c0b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b6f015c4d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79b6f015c4d0>}).
I0420 14:29:53.810839 133857053895808 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0420 14:29:53.810890 133857053895808 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79b6f014aac0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0420 14:29:53.934708 133857053895808 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints
I0420 14:29:53.934956 133857053895808 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_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79b6f099c380>
I0420 14:29:53.935060 133857053895808 checkpointing.py:302] Checkpoint manager created!
I0420 14:29:54.354332 133857053895808 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0420 14:29:54.357341 133857053895808 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.
I0420 14:29:54.409502 133857053895808 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
I0420 14:29:54.432980 133857053895808 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0420 14:29:54.433065 133857053895808 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0420 14:29:55.053699 133857053895808 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_norm_length', 'activation_embed').
I0420 14:29:55.053811 133857053895808 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0420 14:29:55.157567 133857053895808 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 14:29:55.157664 133857053895808 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0420 14:29:55.172443 133857053895808 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 14:29:55.172495 133857053895808 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None).
I0420 14:29:55.193944 133857053895808 attentions.py:1154] attentions/query Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 14:29:55.194016 133857053895808 attentions.py:1154] attentions/query Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0420 14:29:55.208926 133857053895808 attentions.py:1155] attentions/key Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 14:29:55.208978 133857053895808 attentions.py:1155] attentions/key Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0420 14:29:55.223772 133857053895808 attentions.py:1156] attentions/value Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 14:29:55.223822 133857053895808 attentions.py:1156] attentions/value Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0420 14:29:55.245969 133857053895808 attentions.py:1197] attentions/out Logical: bfloat16[8,128,16,128]...................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0420 14:29:55.246039 133857053895808 attentions.py:1197] attentions/out Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None).
I0420 14:29:55.269400 133857053895808 linears.py:525] linears/x Logical: bfloat16[8,128,7168]........................................ ('activation_batch', 'activation_length', 'activation_mlp').
I0420 14:29:55.269459 133857053895808 linears.py:525] linears/x Physical: bfloat16[8,128,7168]........................................ ('fsdp', None, None).
I0420 14:29:55.473196 133857053895808 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0420 14:29:55.473292 133857053895808 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0420 14:29:55.524057 1038858 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
I0420 14:29:55.926017 133857053895808 maxtext_utils.py:1821] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0420 14:29:55.926195 133857053895808 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0420 14:29:55.926239 133857053895808 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0420 14:29:55.926283 133857053895808 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0420 14:29:55.926322 133857053895808 maxtext_utils.py:1821] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0420 14:29:55.926351 133857053895808 maxtext_utils.py:1821] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0420 14:29:55.926389 133857053895808 maxtext_utils.py:1821] 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)
I0420 14:29:55.926429 133857053895808 maxtext_utils.py:1821] 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')
I0420 14:29:55.926457 133857053895808 maxtext_utils.py:1821] 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)
I0420 14:29:55.926485 133857053895808 maxtext_utils.py:1821] 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)
I0420 14:29:55.926519 133857053895808 maxtext_utils.py:1821] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0420 14:29:55.926553 133857053895808 maxtext_utils.py:1821] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0420 14:29:56.387340 133857053895808 train.py:157] train/xent Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0420 14:29:56.387431 133857053895808 train.py:157] train/xent Physical: float32[8,128].............................................. ('fsdp', None).
I0420 14:29:56.401566 133857053895808 train.py:164] train/z_loss Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0420 14:29:56.401618 133857053895808 train.py:164] train/z_loss Physical: float32[8,128].............................................. ('fsdp', None).
I0420 14:30:00.371824 133857053895808 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.
I0420 14:30:00.372395 133857053895808 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0420 14:30:00.372720 133857053895808 metric_logger.py:301] number parameters: 1.104 billion
I0420 14:30:07.920771 133857053895808 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0420 14:30:07.994085 133857053895808 checkpointing.py:798] Waited 0.07328200340270996 seconds for step 0 to finish before starting checkpointing.
I0420 14:30:07.994568 133857053895808 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 14:30:07.994724 133857053895808 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0420 14:30:07.995131 133857053895808 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0.
I0420 14:30:08.086067 133857053895808 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0420 14:30:08.174054 133857053895808 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0420 14:30:08.174150 133857053895808 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
I0420 14:30:08.180672 133742135871040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0
I0420 14:30:08.874331 133857053895808 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.700701s
I0420 14:30:08.874763 133857053895808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 15.679 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 787 milliseconds) (per-host)
I0420 14:30:08.874821 133857053895808 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.787224s (batch_requests_ready=0.082366s, total_serialization_initiated=0.704561s, others=0.000297s)
I0420 14:30:08.874895 133857053895808 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.788259s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.788239)
I0420 14:30:08.875728 133742066665024 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0420 14:30:08.875829 133857053895808 async_checkpointer.py:561] Finished blocking save. Time taken: 0.881066s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0.
I0420 14:30:08.876055 133857053895808 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 14:30:08.876277 133857053895808 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776695407.9945557, 'wait_for_prev_duration_secs': 4.1484832763671875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776695407.9947453, 'checkpointer_blocking_duration_secs': 0.8811967372894287, 'get_old_steps_start_time': 1776695408.8759663, 'get_old_steps_duration_secs': 3.838539123535156e-05, 'checkpoint_manager_blocking_start_time': 1776695407.9944668, 'checkpoint_manager_blocking_duration_secs': 0.8817849159240723}
I0420 14:30:08.876376 133857053895808 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0420 14:30:08.876420 133857053895808 max_utils.py:750]
Memstats: After params initialized:
I0420 14:30:08.876513 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_0(process=0,(0,0,0,0))
I0420 14:30:08.876542 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_1(process=0,(1,0,0,0))
I0420 14:30:08.876562 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_2(process=0,(0,1,0,0))
I0420 14:30:08.876581 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_3(process=0,(1,1,0,0))
I0420 14:30:08.876599 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_4(process=0,(0,2,0,0))
I0420 14:30:08.876617 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_5(process=0,(1,2,0,0))
I0420 14:30:08.876634 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_6(process=0,(0,3,0,0))
I0420 14:30:08.876651 133857053895808 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_7(process=0,(1,3,0,0))
I0420 14:30:08.949815 133742125385280 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/items
I0420 14:30:09.065889 133742087636544 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776695408813566212, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776695408813566212, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0420 14:30:09.066187 133742056179264 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0420 14:30:09.335067 133857053895808 metric_logger.py:196] completed step: 0, seconds: 7.547, TFLOP/s/device: 0.106, Tokens/s/device: 16.959, total_weights: 963, loss: 10.834, lm_loss: 10.834, perplexity: 50699.652
I0420 14:30:09.336244 133857053895808 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/tensorboard/'
I0420 14:30:09.396107 133857053895808 metric_logger.py:196] completed step: 1, seconds: 1.412, TFLOP/s/device: 0.567, Tokens/s/device: 90.671, total_weights: 945, loss: 10.747, lm_loss: 10.747, perplexity: 46511.984
I0420 14:30:09.436849 1042275 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0420 14:30:09.436925 1042275 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0420 14:30:09.460175 133857053895808 metric_logger.py:196] completed step: 2, seconds: 0.011, TFLOP/s/device: 73.529, Tokens/s/device: 11751.744, total_weights: 962, loss: 10.246, lm_loss: 10.246, perplexity: 28171.059
I0420 14:30:09.523889 133857053895808 metric_logger.py:196] completed step: 3, seconds: 0.062, TFLOP/s/device: 12.989, Tokens/s/device: 2076.001, total_weights: 975, loss: 9.829, lm_loss: 9.829, perplexity: 18564.242
I0420 14:30:10.404972 133742114899520 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_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/items/array_metadatas/process_0
I0420 14:30:20.919056 133857053895808 metric_logger.py:196] completed step: 4, seconds: 0.064, TFLOP/s/device: 12.463, Tokens/s/device: 1991.846, total_weights: 962, loss: 9.393, lm_loss: 9.393, perplexity: 12003.683
I0420 14:30:20.928130 133857053895808 metric_logger.py:196] completed step: 5, seconds: 0.064, TFLOP/s/device: 12.602, Tokens/s/device: 2014.162, total_weights: 958, loss: 9.248, lm_loss: 9.248, perplexity: 10381.438
I0420 14:30:20.990583 133857053895808 metric_logger.py:196] completed step: 6, seconds: 11.394, TFLOP/s/device: 0.070, Tokens/s/device: 11.234, total_weights: 930, loss: 9.424, lm_loss: 9.424, perplexity: 12380.232
I0420 14:30:21.054413 133857053895808 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 121.806, Tokens/s/device: 19467.681, total_weights: 935, loss: 9.317, lm_loss: 9.317, perplexity: 11130.819
I0420 14:30:21.118311 133857053895808 metric_logger.py:196] completed step: 8, seconds: 0.063, TFLOP/s/device: 12.661, Tokens/s/device: 2023.587, total_weights: 930, loss: 9.478, lm_loss: 9.478, perplexity: 13071.621
I0420 14:30:21.181529 133857053895808 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0420 14:30:21.182398 133857053895808 checkpointing.py:798] Waited 0.0008952617645263672 seconds for step 9 to finish before starting checkpointing.
I0420 14:30:21.182667 133857053895808 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 14:30:35.636407 133742077150784 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.204822s (commit=25.766605s, array_metadata_write=0.438217s)
I0420 14:30:35.637560 133742066665024 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 458.687 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 27 seconds) (per-host)
I0420 14:30:35.637623 133742066665024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.761749s.
I0420 14:30:35.858811 133742066665024 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776695408813566212, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0420 14:30:36.042704 133742066665024 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0420 14:30:36.248677 133742087636544 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776695408813566212, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/_CHECKPOINT_METADATA
I0420 14:30:36.514272 133742066665024 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0420 14:30:36.514942 133742066665024 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.612002s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/items
I0420 14:30:36.515678 133742066665024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0/items
I0420 14:30:36.765065 133742066665024 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0
I0420 14:30:37.452106 133742066665024 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_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0`.
I0420 14:30:37.452846 133742066665024 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 29.458089s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/0
I0420 14:30:37.452918 133742066665024 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.577044s.
I0420 14:30:37.453037 133742056179264 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 14:30:37.453099 133742056179264 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0420 14:30:37.453154 133742056179264 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0420 14:30:37.453199 133742056179264 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0420 14:30:37.453309 133857053895808 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.
W0420 14:30:37.453456 133857053895808 checkpoint_manager.py:1441] Waiting for previous save to complete took 16.270789 seconds. If this number is high, consider checkpointing less frequently.
I0420 14:30:37.454217 133857053895808 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0420 14:30:37.454528 133857053895808 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9.
I0420 14:30:37.608493 133857053895808 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0420 14:30:37.608597 133857053895808 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
I0420 14:30:37.618074 133742056179264 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9
I0420 14:30:38.324506 133742125385280 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9/items
I0420 14:30:42.000996 133857053895808 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.392958s
I0420 14:30:42.005505 133857053895808 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.755 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host)
I0420 14:30:42.005573 133857053895808 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.478697s (batch_requests_ready=0.078258s, total_serialization_initiated=4.396070s, others=0.004370s)
I0420 14:30:42.005647 133857053895808 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.479235s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=4.479220)
I0420 14:30:42.007332 133742056179264 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0420 14:30:42.007436 133857053895808 async_checkpointer.py:561] Finished blocking save. Time taken: 4.553175s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9.
I0420 14:30:42.007622 133857053895808 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 14:30:42.007780 133742066665024 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0420 14:30:42.007888 133857053895808 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776695421.1826327, 'wait_for_prev_duration_secs': 16.27078938484192, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776695437.4542432, 'checkpointer_blocking_duration_secs': 4.553303003311157, 'get_old_steps_start_time': 1776695442.0075636, 'get_old_steps_duration_secs': 2.1696090698242188e-05, 'checkpoint_manager_blocking_start_time': 1776695421.1825943, 'checkpoint_manager_blocking_duration_secs': 20.825264930725098}
I0420 14:30:42.008012 133857053895808 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0420 14:30:42.008059 133857053895808 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 14:30:42.808142 133742146356800 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_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9/items/array_metadatas/process_0
I0420 14:31:13.981671 133742125385280 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.975365s (commit=31.520880s, array_metadata_write=0.454484s)
I0420 14:31:13.982998 133742056179264 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 346.631 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 36 seconds) (per-host)
I0420 14:31:13.983151 133742056179264 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.975669s.
I0420 14:31:14.427802 133742056179264 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0420 14:31:14.845871 133742056179264 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0420 14:31:14.846573 133742056179264 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.572760s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9/items
I0420 14:31:14.847322 133742056179264 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9/items
I0420 14:31:15.088408 133742056179264 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9
I0420 14:31:15.782211 133742056179264 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_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9`.
I0420 14:31:15.782961 133742056179264 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 38.328706s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_03_dropout/checkpoints/9
I0420 14:31:15.783058 133742056179264 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.775561s.
I0420 14:31:15.783288 133742066665024 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 14:31:15.783415 133742066665024 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0420 14:31:15.783482 133742066665024 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0420 14:31:15.783527 133742066665024 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0420 14:31:15.783712 133857053895808 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.
I0420 14:31:15.783865 133857053895808 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 14:31:15.785026 133857053895808 metric_logger.py:196] completed step: 9, seconds: 0.064, TFLOP/s/device: 12.516, Tokens/s/device: 2000.313, total_weights: 948, loss: 9.141, lm_loss: 9.141, perplexity: 9333.774
Per train step:
Total TFLOPs: 0.80
split as 99.60% learnable weight flops and 0.40% attention flops