2026-04-16 12:45:42.439497: 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 12:45:42.558484 132015521307776 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 12:46:35.286261 132015521307776 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 12:46:35.286366 132015521307776 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 12:46:35.286400 132015521307776 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 12:46:35.286428 132015521307776 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system
I0416 12:46:35.286501 132015521307776 train.py:683] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 12:46:35.376817 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 12:46:35.773447 132015521307776 maxtext_utils.py:1517] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 12:46:35.773914 132015521307776 checkpointing.py:677] Setting up checkpoint logger...
I0416 12:46:35.773961 132015521307776 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 12:46:35.774000 132015521307776 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 12:46:35.775263 132015521307776 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 0x7810ac9a2e40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 12:46:38.477303 132015521307776 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0416 12:46:38.477470 132015521307776 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 0x780bdfef0d40>}, handler_registry=None
I0416 12:46:38.477732 132015521307776 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x780bdfef0d40>` 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 12:46:38.477773 132015521307776 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x780a2f342090>` 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 12:46:38.477804 132015521307776 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 0x780bdfef0d40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x780bdfef0d40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x780a2f342090>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x780a2f342090>}).
I0416 12:46:38.478367 132015521307776 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 12:46:38.478423 132015521307776 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x780a2f338ea0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 12:46:38.613216 132015521307776 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints
I0416 12:46:38.613427 132015521307776 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x780a2f341790>
I0416 12:46:38.613507 132015521307776 checkpointing.py:301] Checkpoint manager created!
I0416 12:46:38.694790 132015521307776 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0416 12:46:38.697795 132015521307776 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 12:46:38.751123 132015521307776 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 12:46:38.779544 132015521307776 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 12:46:38.779610 132015521307776 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 12:46:39.926918 132015521307776 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 12:46:39.927038 132015521307776 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 12:46:40.032032 132015521307776 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 12:46:40.032119 132015521307776 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 12:46:40.047012 132015521307776 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 12:46:40.047071 132015521307776 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 12:46:40.122297 132015521307776 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 12:46:40.122432 132015521307776 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 12:46:40.137454 132015521307776 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 12:46:40.137501 132015521307776 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 12:46:40.152511 132015521307776 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0416 12:46:40.152561 132015521307776 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 12:46:40.174754 132015521307776 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 12:46:40.174810 132015521307776 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 12:46:40.231898 132015521307776 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 12:46:40.231965 132015521307776 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 12:46:40.666742 132015521307776 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 12:46:40.666832 132015521307776 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
[DECOUPLED NO-OP] gcs_storage: using stubs.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] mldiagnostics: using stub.
[DECOUPLED NO-OP] workload_monitor: using stub.
[DECOUPLED NO-OP] vertex_tensorboard: using stub.
fsdp: 8
I0416 12:46:42.582001 132015521307776 maxtext_utils.py:1620] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 12:46:42.582126 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 12:46:42.582172 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0416 12:46:42.582219 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0416 12:46:42.582260 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 12:46:42.582288 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0416 12:46:42.582327 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/key/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0416 12:46:42.582364 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/out/kernel
Shape: float32[16,16,128,2048]
Logical: P('heads', 'layers', 'kv', 'embed')
Physical: (None, None, None, 'fsdp')
I0416 12:46:42.582391 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/query/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'q_heads', 'kv')
Physical: ('fsdp', None, None, None)
I0416 12:46:42.582417 132015521307776 maxtext_utils.py:1620] params/params/decoder/layers/self_attention/value/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0416 12:46:42.582450 132015521307776 maxtext_utils.py:1620] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 12:46:42.582484 132015521307776 maxtext_utils.py:1620] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 12:46:44.247925 132015521307776 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 12:46:44.248015 132015521307776 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 12:46:44.262416 132015521307776 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 12:46:44.262466 132015521307776 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 12:46:44.806258 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 12:46:45.000869 132015521307776 max_utils.py:791] Total memory size: 3.7 GB, Output size: 1.5 GB, Temp size: 2.2 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0416 12:46:45.001472 132015521307776 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 12:46:45.001619 132015521307776 metric_logger.py:289] number parameters: 1.104 billion
W0416 12:46:49.482879 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 12:46:49.705065 132015521307776 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0416 12:46:49.829895 132015521307776 checkpointing.py:776] Waited 0.12482500076293945 seconds for step 0 to finish before starting checkpointing.
I0416 12:46:49.830346 132015521307776 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 12:46:49.830514 132015521307776 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 12:46:49.830884 132015521307776 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0.
I0416 12:46:49.928123 132015521307776 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 12:46:50.023104 131900242921024 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0
I0416 12:46:50.023754 132015521307776 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 12:46:50.023917 132015521307776 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 12:46:50.259703 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.274158 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.289011 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.304728 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.317932 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.324197 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.328676 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.333021 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 12:46:50.593346 132015521307776 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.570510s
I0416 12:46:50.593789 132015521307776 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 18.574 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 664 milliseconds) (per-host)
I0416 12:46:50.593840 132015521307776 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.664512s (batch_requests_ready=0.090142s, total_serialization_initiated=0.574027s, others=0.000344s)
I0416 12:46:50.593904 132015521307776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.665326s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.665306)
I0416 12:46:50.594727 131900175812160 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 12:46:50.594815 132015521307776 async_checkpointer.py:561] Finished blocking save. Time taken: 0.764261s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0.
I0416 12:46:50.594982 132015521307776 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 12:46:50.595220 131900253406784 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 12:46:50.595329 132015521307776 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776343609.8303335, 'wait_for_prev_duration_secs': 4.291534423828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776343609.8305368, 'checkpointer_blocking_duration_secs': 0.7643752098083496, 'get_old_steps_start_time': 1776343610.594928, 'get_old_steps_duration_secs': 2.0265579223632812e-05, 'checkpoint_manager_blocking_start_time': 1776343609.8302512, 'checkpoint_manager_blocking_duration_secs': 0.7650468349456787}
I0416 12:46:50.595436 132015521307776 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0416 12:46:50.595473 132015521307776 max_utils.py:750]
Memstats: After params initialized:
I0416 12:46:50.595578 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0))
I0416 12:46:50.595604 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_1(process=0,(1,0,0,0))
I0416 12:46:50.595624 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_2(process=0,(0,1,0,0))
I0416 12:46:50.595642 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_3(process=0,(1,1,0,0))
I0416 12:46:50.595659 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_4(process=0,(0,2,0,0))
I0416 12:46:50.595681 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_5(process=0,(1,2,0,0))
I0416 12:46:50.595697 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_6(process=0,(0,3,0,0))
I0416 12:46:50.595713 132015521307776 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_7(process=0,(1,3,0,0))
W0416 12:46:50.601593 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.607234 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 12:46:50.613112 3828239 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 12:46:50.736939 131900232435264 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/items
I0416 12:46:50.931341 131900156937792 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343610645505120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776343610645505120, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/_CHECKPOINT_METADATA
I0416 12:46:50.940893 132015521307776 metric_logger.py:185] completed step: 0, seconds: 4.703, TFLOP/s/device: 2.889, Tokens/s/device: 435.470, total_weights: 13328, loss: 10.840
I0416 12:46:50.941761 132015521307776 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/tensorboard/'
I0416 12:46:51.061065 132015521307776 metric_logger.py:185] completed step: 1, seconds: 1.234, TFLOP/s/device: 11.011, Tokens/s/device: 1659.712, total_weights: 12332, loss: 10.869
I0416 12:46:51.182417 132015521307776 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 1410.038, Tokens/s/device: 212536.322, total_weights: 15161, loss: 9.874
I0416 12:46:51.304309 132015521307776 metric_logger.py:185] completed step: 3, seconds: 0.122, TFLOP/s/device: 111.766, Tokens/s/device: 16846.539, total_weights: 13327, loss: 9.406
I0416 12:46:51.440439 3831346 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 12:46:51.440498 3831346 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 12:46:52.442063 131900196783680 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/items/array_metadatas/process_0
I0416 12:47:18.069374 131900186297920 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.634531s (commit=26.164198s, array_metadata_write=0.470332s)
I0416 12:47:18.070611 131900175812160 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 449.049 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 28 seconds) (per-host)
I0416 12:47:18.070715 131900175812160 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 27.475855s.
I0416 12:47:18.282022 131900175812160 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343610645505120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/_CHECKPOINT_METADATA
I0416 12:47:18.468897 131900175812160 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 12:47:18.653415 131900156937792 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776343610645505120, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/_CHECKPOINT_METADATA
I0416 12:47:18.908901 131900175812160 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 12:47:18.909457 131900175812160 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.584264s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/items
I0416 12:47:18.910024 131900175812160 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0/items
I0416 12:47:19.159909 131900175812160 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0
I0416 12:47:19.828723 131900175812160 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0`.
I0416 12:47:19.829377 131900175812160 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 29.998829s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/0
I0416 12:47:19.829462 131900175812160 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 29.234603s.
I0416 12:47:19.829646 131900253406784 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 12:47:19.829757 131900253406784 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 12:47:19.829834 131900253406784 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 12:47:19.829885 131900253406784 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 12:47:20.445640 132015521307776 metric_logger.py:185] completed step: 4, seconds: 0.121, TFLOP/s/device: 112.275, Tokens/s/device: 16923.382, total_weights: 11939, loss: 9.149
I0416 12:47:20.455201 132015521307776 metric_logger.py:185] completed step: 5, seconds: 0.122, TFLOP/s/device: 111.255, Tokens/s/device: 16769.566, total_weights: 15502, loss: 9.058
I0416 12:47:20.575615 132015521307776 metric_logger.py:185] completed step: 6, seconds: 29.141, TFLOP/s/device: 0.466, Tokens/s/device: 70.280, total_weights: 13864, loss: 8.892
I0416 12:47:20.697926 132015521307776 metric_logger.py:185] completed step: 7, seconds: 0.007, TFLOP/s/device: 2077.860, Tokens/s/device: 313197.737, total_weights: 12988, loss: 8.751
I0416 12:47:20.819894 132015521307776 metric_logger.py:185] completed step: 8, seconds: 0.121, TFLOP/s/device: 111.983, Tokens/s/device: 16879.306, total_weights: 13820, loss: 8.791
I0416 12:47:20.941859 132015521307776 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0416 12:47:20.942699 132015521307776 checkpointing.py:776] Waited 0.0008695125579833984 seconds for step 9 to finish before starting checkpointing.
I0416 12:47:20.942964 132015521307776 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 12:47:20.943142 132015521307776 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 12:47:20.943429 132015521307776 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9.
I0416 12:47:21.113486 132015521307776 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 12:47:21.113624 131900253406784 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9
I0416 12:47:21.113703 132015521307776 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 12:47:21.839188 131900232435264 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9/items
I0416 12:47:23.454983 132015521307776 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.342227s
I0416 12:47:23.459132 132015521307776 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.076 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host)
I0416 12:47:23.459191 132015521307776 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.431199s (batch_requests_ready=0.081814s, total_serialization_initiated=2.345349s, others=0.004036s)
I0416 12:47:23.459259 132015521307776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.431707s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=2.431689)
I0416 12:47:23.460791 131900186297920 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 12:47:23.460887 132015521307776 async_checkpointer.py:561] Finished blocking save. Time taken: 2.517703s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9.
I0416 12:47:23.461087 132015521307776 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 12:47:23.461264 131900175812160 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 12:47:23.461372 132015521307776 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776343640.9429252, 'wait_for_prev_duration_secs': 6.532669067382812e-05, 'time_between_consecutive_saves_sec': 1.1130187511444092, 'checkpointer_blocking_start_time': 1776343640.9431653, 'checkpointer_blocking_duration_secs': 2.5178275108337402, 'get_old_steps_start_time': 1776343643.461008, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776343640.9428902, 'checkpoint_manager_blocking_duration_secs': 2.518458604812622}
I0416 12:47:23.461490 132015521307776 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0416 12:47:23.461524 132015521307776 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 12:47:24.177363 131900221949504 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9/items/array_metadatas/process_0
I0416 12:47:59.442591 131900232435264 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.982671s (commit=35.506034s, array_metadata_write=0.476636s)
I0416 12:47:59.443850 131900186297920 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 328.948 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host)
I0416 12:47:59.443961 131900186297920 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.983029s.
I0416 12:47:59.879284 131900186297920 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 12:48:00.366701 131900186297920 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 12:48:00.367287 131900186297920 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.619087s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9/items
I0416 12:48:00.368029 131900186297920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9/items
I0416 12:48:00.618110 131900186297920 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9
I0416 12:48:01.309664 131900186297920 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9`.
I0416 12:48:01.310295 131900186297920 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.367116s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260416_123638/linen_main_20260416_123638_04_int8/checkpoints/9
I0416 12:48:01.310383 131900186297920 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.849453s.
I0416 12:48:01.310611 131900175812160 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 12:48:01.310721 131900175812160 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 12:48:01.310789 131900175812160 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 12:48:01.310838 131900175812160 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 12:48:01.311403 132015521307776 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 12:48:01.312119 132015521307776 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 12:48:01.313089 132015521307776 metric_logger.py:185] completed step: 9, seconds: 0.122, TFLOP/s/device: 111.232, Tokens/s/device: 16766.134, total_weights: 12300, loss: 8.774
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops