2026-04-15 21:26:36.170286: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)
I0415 21:26:36.289074 140340394753152 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0415 21:27:22.786390 140340394753152 max_utils.py:800] System Information: Jax Version: 0.9.2
I0415 21:27:22.786495 140340394753152 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0415 21:27:22.786529 140340394753152 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894
I0415 21:27:22.786567 140340394753152 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0415 21:27:22.786646 140340394753152 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0415 21:27:22.877105 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:23.262890 140340394753152 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0415 21:27:23.263170 140340394753152 checkpointing.py:677] Setting up checkpoint logger...
I0415 21:27:23.263215 140340394753152 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0415 21:27:23.263255 140340394753152 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0415 21:27:23.263709 140340394753152 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 0x7fa2f47fbb00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0415 21:27:25.631195 140340394753152 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0415 21:27:25.631366 140340394753152 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 0x7f9c7853dc40>}, handler_registry=None
I0415 21:27:25.631638 140340394753152 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9c7853dc40>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`.
I0415 21:27:25.631677 140340394753152 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9c7853e210>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`.
I0415 21:27:25.631707 140340394753152 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 0x7f9c7853dc40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9c7853dc40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9c7853e210>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9c7853e210>}).
I0415 21:27:25.632123 140340394753152 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0415 21:27:25.632177 140340394753152 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7f9c78534fe0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0415 21:27:25.760568 140340394753152 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints
I0415 21:27:25.760792 140340394753152 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f9e29069580>
I0415 21:27:25.760871 140340394753152 checkpointing.py:301] Checkpoint manager created!
I0415 21:27:25.840600 140340394753152 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0415 21:27:25.843567 140340394753152 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.
I0415 21:27:25.897167 140340394753152 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
I0415 21:27:25.925784 140340394753152 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0415 21:27:25.925849 140340394753152 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0415 21:27:26.469923 140340394753152 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0415 21:27:26.471690 140340394753152 reader.py:262] Creating a tf.data.Dataset reading 2 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0.
I0415 21:27:26.487701 140340394753152 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split validation, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0415 21:27:26.490296 140340394753152 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0415 21:27:26.490344 140340394753152 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0415 21:27:27.182516 140340394753152 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0415 21:27:27.182614 140340394753152 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:27:27.283853 140340394753152 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:27:27.283929 140340394753152 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:27:27.299035 140340394753152 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0415 21:27:27.299094 140340394753152 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0415 21:27:27.320806 140340394753152 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:27:27.320859 140340394753152 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:27:27.336081 140340394753152 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:27:27.336131 140340394753152 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:27:27.351325 140340394753152 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0415 21:27:27.351374 140340394753152 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:27:27.374838 140340394753152 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0415 21:27:27.374896 140340394753152 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0415 21:27:27.394079 140340394753152 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0415 21:27:27.394130 140340394753152 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0415 21:27:27.581523 140340394753152 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0415 21:27:27.581617 140340394753152 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
W0415 21:27:27.632250 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
[DECOUPLED NO-OP] gcs_storage: using stubs.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] workload_monitor: using stub.
[DECOUPLED NO-OP] vertex_tensorboard: using stub.
fsdp: 8
I0415 21:27:28.026564 140340394753152 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0415 21:27:28.026672 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0415 21:27:28.026713 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0415 21:27:28.026757 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0415 21:27:28.026796 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0415 21:27:28.026826 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0415 21:27:28.026865 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/key/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0415 21:27:28.026911 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/out/kernel
Shape: float32[16,16,128,2048]
Logical: P('heads', 'layers', 'kv', 'embed')
Physical: (None, None, None, 'fsdp')
I0415 21:27:28.026942 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/query/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'q_heads', 'kv')
Physical: ('fsdp', None, None, None)
I0415 21:27:28.026970 140340394753152 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/value/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0415 21:27:28.027005 140340394753152 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0415 21:27:28.027039 140340394753152 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0415 21:27:28.459878 140340394753152 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:27:28.459961 140340394753152 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0415 21:27:28.474469 140340394753152 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0415 21:27:28.474519 140340394753152 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0415 21:27:28.748417 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:28.864385 140340394753152 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0415 21:27:28.864861 140340394753152 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0415 21:27:28.865017 140340394753152 metric_logger.py:289] number parameters: 1.104 billion
W0415 21:27:31.896891 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:32.028528 140340394753152 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0415 21:27:32.148121 140340394753152 checkpointing.py:776] Waited 0.11958479881286621 seconds for step 0 to finish before starting checkpointing.
I0415 21:27:32.148552 140340394753152 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:27:32.148710 140340394753152 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0415 21:27:32.149090 140340394753152 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0.
I0415 21:27:32.253256 140340394753152 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0415 21:27:32.343091 140340394753152 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:27:32.343181 140340394753152 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0415 21:27:32.354951 140225330284096 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0
W0415 21:27:32.557273 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.572480 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.581281 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.596529 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.608376 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.612769 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.617516 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.621658 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:32.978262 140340394753152 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.635627s
I0415 21:27:32.978707 140340394753152 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 17.041 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 724 milliseconds) (per-host)
I0415 21:27:32.978758 140340394753152 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.724280s (batch_requests_ready=0.084801s, total_serialization_initiated=0.639130s, others=0.000349s)
I0415 21:27:32.978817 140340394753152 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.725100s (all_items=0.000020s, per_item={'items': '0.00001979'}, temp_paths=0.725080)
I0415 21:27:32.979601 140225258980928 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:27:32.979694 140340394753152 async_checkpointer.py:561] Finished blocking save. Time taken: 0.830945s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0.
I0415 21:27:32.979861 140340394753152 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:27:32.980022 140225340769856 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:27:32.980137 140340394753152 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288452.1485388, 'wait_for_prev_duration_secs': 4.100799560546875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288452.1487317, 'checkpointer_blocking_duration_secs': 0.8310589790344238, 'get_old_steps_start_time': 1776288452.9798086, 'get_old_steps_duration_secs': 2.002716064453125e-05, 'checkpoint_manager_blocking_start_time': 1776288452.1484554, 'checkpoint_manager_blocking_duration_secs': 0.8316521644592285}
I0415 21:27:32.980245 140340394753152 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0415 21:27:32.980282 140340394753152 max_utils.py:750]
Memstats: After params initialized:
I0415 21:27:32.980355 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0415 21:27:32.980382 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0415 21:27:32.980404 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0415 21:27:32.980421 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0415 21:27:32.980439 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0415 21:27:32.980457 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0415 21:27:32.980474 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0415 21:27:32.980492 140340394753152 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0415 21:27:32.986294 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.991992 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0415 21:27:32.997769 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:33.103455 140225319798336 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/items
I0415 21:27:33.320154 140225248495168 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288453006749901, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776288453006749901, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:27:33.329032 140340394753152 metric_logger.py:185] completed step: 0, seconds: 3.163, TFLOP/s/device: 4.296, Tokens/s/device: 647.484, total_weights: 13328, loss: 10.828
I0415 21:27:33.329924 140340394753152 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/tensorboard/'
I0415 21:27:33.444714 140340394753152 metric_logger.py:185] completed step: 1, seconds: 1.299, TFLOP/s/device: 10.463, Tokens/s/device: 1577.145, total_weights: 12332, loss: 10.828
I0415 21:27:33.562212 140340394753152 metric_logger.py:185] completed step: 2, seconds: 0.009, TFLOP/s/device: 1483.149, Tokens/s/device: 223556.380, total_weights: 15161, loss: 9.823
I0415 21:27:34.192120 3245423 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0415 21:27:34.192175 3245423 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0415 21:27:35.072754 140225279952448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/items/array_metadatas/process_0
W0415 21:27:40.582293 3242174 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0415 21:27:40.680166 140340394753152 train.py:600] Completed eval step 0
I0415 21:27:40.721447 140340394753152 train.py:600] Completed eval step 1
I0415 21:27:40.725900 140340394753152 metric_logger.py:204] eval metrics after step: 4, loss=8.979, total_weights=26703.0
I0415 21:27:40.726263 140340394753152 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.302, Tokens/s/device: 17530.344, total_weights: 13327, loss: 9.405
I0415 21:28:00.391901 140340394753152 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.499, Tokens/s/device: 17409.193, total_weights: 11939, loss: 9.045
I0415 21:28:00.401089 140340394753152 metric_logger.py:185] completed step: 5, seconds: 7.163, TFLOP/s/device: 1.897, Tokens/s/device: 285.899, total_weights: 15502, loss: 8.953
I0415 21:28:00.518031 140340394753152 metric_logger.py:185] completed step: 6, seconds: 19.666, TFLOP/s/device: 0.691, Tokens/s/device: 104.139, total_weights: 13864, loss: 8.775
I0415 21:28:00.635414 140340394753152 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2269.818, Tokens/s/device: 342131.640, total_weights: 12988, loss: 8.688
I0415 21:28:01.308006 140340394753152 train.py:600] Completed eval step 0
I0415 21:28:01.346355 140340394753152 train.py:600] Completed eval step 1
I0415 21:28:01.349855 140340394753152 metric_logger.py:204] eval metrics after step: 9, loss=8.653, total_weights=26703.0
I0415 21:28:01.350195 140340394753152 metric_logger.py:185] completed step: 8, seconds: 0.118, TFLOP/s/device: 115.337, Tokens/s/device: 17384.809, total_weights: 13820, loss: 8.740
I0415 21:28:01.352163 140340394753152 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0415 21:28:01.352757 140340394753152 checkpointing.py:776] Waited 0.0005979537963867188 seconds for step 9 to finish before starting checkpointing.
I0415 21:28:01.352993 140340394753152 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:28:05.661067 140225269466688 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.476167s (commit=31.037598s, array_metadata_write=0.438569s)
I0415 21:28:05.662063 140225258980928 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 378.261 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host)
I0415 21:28:05.662120 140225258980928 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.682383s.
I0415 21:28:05.954188 140225258980928 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288453006749901, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:28:06.149986 140225258980928 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0415 21:28:06.353561 140225248495168 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776288453006749901, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0415 21:28:06.583543 140225258980928 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:28:06.584200 140225258980928 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.590499s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/items
I0415 21:28:06.584888 140225258980928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0/items
I0415 21:28:06.827769 140225258980928 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0
I0415 21:28:07.516530 140225258980928 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0`.
I0415 21:28:07.517167 140225258980928 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.368423s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/0
I0415 21:28:07.517247 140225258980928 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 34.537511s.
I0415 21:28:07.517429 140225340769856 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:28:07.517541 140225340769856 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:28:07.517613 140225340769856 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:28:07.517660 140225340769856 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0415 21:28:07.517813 140340394753152 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0415 21:28:07.517946 140340394753152 checkpoint_manager.py:1441] Waiting for previous save to complete took 6.164952 seconds. If this number is high, consider checkpointing less frequently.
I0415 21:28:07.518793 140340394753152 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0415 21:28:07.519099 140340394753152 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9.
I0415 21:28:07.701242 140225340769856 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9
I0415 21:28:07.716638 140340394753152 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0415 21:28:07.716730 140340394753152 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0415 21:28:08.429952 140224388662848 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9/items
I0415 21:28:14.333070 140340394753152 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 6.616843s
I0415 21:28:14.338253 140340394753152 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.835 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host)
I0415 21:28:14.338322 140340394753152 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 6.726924s (batch_requests_ready=0.101952s, total_serialization_initiated=6.619883s, others=0.005089s)
I0415 21:28:14.338398 140340394753152 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 6.727453s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=6.727435)
I0415 21:28:14.340109 140225340769856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0415 21:28:14.340221 140340394753152 async_checkpointer.py:561] Finished blocking save. Time taken: 6.821371s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9.
I0415 21:28:14.340473 140340394753152 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0415 21:28:14.340637 140225258980928 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0415 21:28:14.340744 140340394753152 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776288481.3529603, 'wait_for_prev_duration_secs': 6.164951801300049, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776288487.5188181, 'checkpointer_blocking_duration_secs': 6.821548223495483, 'get_old_steps_start_time': 1776288494.34039, 'get_old_steps_duration_secs': 3.4809112548828125e-05, 'checkpoint_manager_blocking_start_time': 1776288481.352919, 'checkpoint_manager_blocking_duration_secs': 12.987797260284424}
I0415 21:28:14.340868 140340394753152 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0415 21:28:14.340903 140340394753152 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0415 21:28:15.159859 140225279952448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9/items/array_metadatas/process_0
I0415 21:28:52.004823 140224388662848 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.665694s (commit=37.170302s, array_metadata_write=0.495392s)
I0415 21:28:52.006094 140225340769856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 284.647 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host)
I0415 21:28:52.006213 140225340769856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.665932s.
I0415 21:28:52.443566 140225340769856 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0.
I0415 21:28:52.875964 140225340769856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0415 21:28:52.876782 140225340769856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.575953s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9/items
I0415 21:28:52.877427 140225340769856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9/items
I0415 21:28:53.118739 140225340769856 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9
I0415 21:28:53.816416 140225340769856 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9`.
I0415 21:28:53.817069 140225340769856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.298236s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260415_211126/linen_main_20260415_211126_07_eval/checkpoints/9
I0415 21:28:53.817149 140225340769856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.476871s.
I0415 21:28:53.817351 140225258980928 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0415 21:28:53.817453 140225258980928 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0415 21:28:53.817540 140225258980928 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0415 21:28:53.817595 140225258980928 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0415 21:28:53.818138 140340394753152 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0415 21:28:53.818831 140340394753152 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0415 21:28:53.819801 140340394753152 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 115.969, Tokens/s/device: 17480.070, total_weights: 12300, loss: 8.710
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops