2026-04-22 17:59:55.069109: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776880795.082792 2032837 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776880795.086739 2032837 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776880795.098381 2032837 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880795.098397 2032837 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880795.098399 2032837 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880795.098400 2032837 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-22 17:59:58.761073: 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) I0422 17:59:58.896213 140634764187456 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 18:00:31.129326 140634764187456 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 18:00:31.129589 140634764187456 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 18:00:31.129640 140634764187456 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0422 18:00:31.129669 140634764187456 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 18:00:31.129693 140634764187456 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 18:00:31.129868 140634764187456 train.py:706] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 18:00:31.779716 140634764187456 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 18:00:31.779976 140634764187456 checkpointing.py:677] Setting up checkpoint logger... I0422 18:00:31.780023 140634764187456 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 18:00:31.780073 140634764187456 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 18:00:31.780495 140634764187456 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 0x7fe7cb9bee40>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 18:00:34.278924 140634764187456 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 18:00:34.279320 140634764187456 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 0x7fe5f8235e50>}, handler_registry=None I0422 18:00:34.279556 140634764187456 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe5f8235e50>` 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`. I0422 18:00:34.279597 140634764187456 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe5f8234bf0>` 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`. I0422 18:00:34.279626 140634764187456 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 0x7fe5f8235e50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fe5f8235e50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe5f8234bf0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fe5f8234bf0>}). I0422 18:00:34.279905 140634764187456 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 18:00:34.279958 140634764187456 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7fe5f8254900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 18:00:34.407696 140634764187456 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints I0422 18:00:34.407931 140634764187456 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_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fe5f8234770> I0422 18:00:34.408229 140634764187456 checkpointing.py:301] Checkpoint manager created! I0422 18:00:35.333272 140634764187456 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 18:00:35.333370 140634764187456 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:00:35.613663 140634764187456 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:00:35.613757 140634764187456 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:00:35.626054 140634764187456 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 18:00:35.626104 140634764187456 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 18:00:35.645184 140634764187456 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:00:35.645244 140634764187456 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:00:35.657596 140634764187456 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:00:35.657650 140634764187456 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:00:35.669888 140634764187456 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 18:00:35.669936 140634764187456 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:00:35.689366 140634764187456 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 18:00:35.689429 140634764187456 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 18:00:35.704875 140634764187456 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 18:00:35.704930 140634764187456 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 18:00:35.855094 140634764187456 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 18:00:35.855195 140634764187456 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 I0422 18:00:36.470752 140634764187456 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 18:00:36.470878 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 18:00:36.470921 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 18:00:36.470968 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0422 18:00:36.471010 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 18:00:36.471039 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 18:00:36.471096 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 18:00:36.471135 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0422 18:00:36.471164 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0422 18:00:36.471190 140634764187456 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 18:00:36.471224 140634764187456 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 18:00:36.471259 140634764187456 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 18:00:36.488089 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0422 18:00:36.488149 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0422 18:00:36.499223 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0422 18:00:36.499270 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0422 18:00:36.521323 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0422 18:00:36.521378 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0422 18:00:36.532384 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0422 18:00:36.532447 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0422 18:00:36.554412 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0422 18:00:36.554471 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0422 18:00:36.565462 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0422 18:00:36.565511 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0422 18:00:36.598413 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0422 18:00:36.598482 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0422 18:00:36.609476 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0422 18:00:36.609527 140634764187456 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0422 18:00:37.139029 140634764187456 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:00:37.139132 140634764187456 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:00:37.151016 140634764187456 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 18:00:37.151177 140634764187456 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 18:00:45.375000 140634764187456 max_utils.py:791] Total memory size: 4.5 GB, Output size: 1.5 GB, Temp size: 2.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0422 18:00:45.375984 140634764187456 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 18:00:45.376355 140634764187456 metric_logger.py:301] number parameters: 1.104 billion I0422 18:00:47.962504 140634764187456 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 18:00:48.396699 140634764187456 checkpointing.py:776] Waited 0.434171199798584 seconds for step 0 to finish before starting checkpointing. I0422 18:00:48.397388 140634764187456 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:00:48.397564 140634764187456 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 18:00:48.397998 140634764187456 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0. I0422 18:00:48.484704 140634764187456 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 18:00:48.572616 140603440498240 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0 I0422 18:00:48.581214 140634764187456 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:00:48.581311 140634764187456 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 I0422 18:00:49.392986 140603430012480 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/items I0422 18:00:49.545391 140603222394432 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880849298303094, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776880849298303094, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0422 18:00:49.911772 2035232 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 18:00:49.911827 2035232 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 18:00:51.876227 140634764187456 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.295522s I0422 18:00:51.881224 140634764187456 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.635 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 18:00:51.881291 140634764187456 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.395166s (batch_requests_ready=0.091154s, total_serialization_initiated=3.299133s, others=0.004879s) I0422 18:00:51.881371 140634764187456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.396100s (all_items=0.000032s, per_item={'items': '0.00003242'}, temp_paths=3.396067) I0422 18:00:51.882976 140603635533376 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:00:51.883092 140634764187456 async_checkpointer.py:561] Finished blocking save. Time taken: 3.485469s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0. I0422 18:00:51.883282 140634764187456 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:00:51.883449 140602417088064 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:00:51.883545 140634764187456 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880848.3973756, 'wait_for_prev_duration_secs': 4.076957702636719e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880848.3976052, 'checkpointer_blocking_duration_secs': 3.485600233078003, 'get_old_steps_start_time': 1776880851.8832226, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776880848.3972957, 'checkpoint_manager_blocking_duration_secs': 3.4862239360809326} I0422 18:00:51.883656 140634764187456 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 18:00:51.883695 140634764187456 max_utils.py:750] Memstats: After params initialized: I0422 18:00:51.883743 140634764187456 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0422 18:00:51.883773 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0422 18:00:51.883794 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0422 18:00:51.883812 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0422 18:00:51.883829 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0422 18:00:51.883846 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0422 18:00:51.883861 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0422 18:00:51.883877 140634764187456 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0422 18:00:52.268738 140634764187456 metric_logger.py:196] completed step: 0, seconds: 2.585, TFLOP/s/device: 21.022, Tokens/s/device: 3168.664, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0422 18:00:52.269759 140634764187456 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/tensorboard/' I0422 18:00:52.695609 140634764187456 metric_logger.py:196] completed step: 1, seconds: 4.304, TFLOP/s/device: 12.627, Tokens/s/device: 1903.333, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0422 18:00:53.119442 140634764187456 metric_logger.py:196] completed step: 2, seconds: 0.007, TFLOP/s/device: 7328.549, Tokens/s/device: 1104638.619, total_weights: 65536, loss: 10.261, lm_loss: 10.261, perplexity: 28586.895 I0422 18:00:53.124088 140603268531776 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/items/array_metadatas/process_0 I0422 18:00:53.544528 140634764187456 metric_logger.py:196] completed step: 3, seconds: 0.428, TFLOP/s/device: 127.028, Tokens/s/device: 19146.987, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16845.295 I0422 18:01:27.698460 140634764187456 metric_logger.py:196] completed step: 4, seconds: 0.426, TFLOP/s/device: 127.559, Tokens/s/device: 19227.023, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10654.189 I0422 18:01:27.708383 140634764187456 metric_logger.py:196] completed step: 5, seconds: 0.426, TFLOP/s/device: 127.572, Tokens/s/device: 19229.099, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7241.146 I0422 18:01:28.131392 140634764187456 metric_logger.py:196] completed step: 6, seconds: 34.153, TFLOP/s/device: 1.591, Tokens/s/device: 239.861, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5368.870 I0422 18:01:28.554974 140634764187456 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 9033.995, Tokens/s/device: 1361702.128, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4361.125 I0422 18:01:28.979547 140634764187456 metric_logger.py:196] completed step: 8, seconds: 0.423, TFLOP/s/device: 128.355, Tokens/s/device: 19346.992, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3832.059 I0422 18:01:29.402942 140634764187456 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0422 18:01:29.404129 140634764187456 checkpointing.py:776] Waited 0.0012090206146240234 seconds for step 9 to finish before starting checkpointing. I0422 18:01:29.404432 140634764187456 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:01:33.041108 140602610026048 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.159051s (commit=40.683571s, array_metadata_write=0.475480s) I0422 18:01:33.042366 140603635533376 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 283.615 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0422 18:01:33.042484 140603635533376 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.159346s. I0422 18:01:33.311590 140603635533376 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880849298303094, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0422 18:01:33.531337 140603635533376 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 18:01:33.731910 140603222394432 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880849298303094, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0422 18:01:33.980837 140603635533376 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:01:33.981024 140603635533376 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.627135s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/items I0422 18:01:33.981784 140603635533376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0/items I0422 18:01:34.226526 140603635533376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0 I0422 18:01:34.948902 140603635533376 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0`. I0422 18:01:34.949640 140603635533376 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.552022s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/0 I0422 18:01:34.949717 140603635533376 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 43.066582s. I0422 18:01:34.949949 140602417088064 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:01:34.950089 140602417088064 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:01:34.950148 140602417088064 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:01:34.950195 140602417088064 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 18:01:34.950330 140634764187456 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. W0422 18:01:34.950471 140634764187456 checkpoint_manager.py:1441] Waiting for previous save to complete took 5.546057 seconds. If this number is high, consider checkpointing less frequently. I0422 18:01:34.951315 140634764187456 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0422 18:01:34.951612 140634764187456 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9. I0422 18:01:35.143588 140634764187456 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0422 18:01:35.143691 140634764187456 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 I0422 18:01:35.144733 140603635533376 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9 I0422 18:01:35.877403 140603268531776 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9/items I0422 18:01:36.756293 140634764187456 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.613313s I0422 18:01:36.760724 140634764187456 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 7.215 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0422 18:01:36.760796 140634764187456 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.710588s (batch_requests_ready=0.089838s, total_serialization_initiated=1.616430s, others=0.004319s) I0422 18:01:36.760872 140634764187456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.711178s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=1.711161) I0422 18:01:36.762767 140603635533376 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 18:01:36.762884 140634764187456 async_checkpointer.py:561] Finished blocking save. Time taken: 1.811526s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9. I0422 18:01:36.763034 140634764187456 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 18:01:36.763240 140602417088064 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 18:01:36.763350 140634764187456 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880889.4043753, 'wait_for_prev_duration_secs': 5.546057224273682, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880894.9513383, 'checkpointer_blocking_duration_secs': 1.8116073608398438, 'get_old_steps_start_time': 1776880896.7629638, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776880889.4043367, 'checkpoint_manager_blocking_duration_secs': 7.358988046646118} I0422 18:01:36.763461 140634764187456 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0422 18:01:36.763500 140634764187456 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 18:01:37.518085 140603258046016 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9/items/array_metadatas/process_0 I0422 18:02:14.556259 140603268531776 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.794343s (commit=37.355464s, array_metadata_write=0.438879s) I0422 18:02:14.557361 140603635533376 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 319.861 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0422 18:02:14.557418 140603635533376 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.794568s. I0422 18:02:15.044504 140603635533376 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 18:02:15.468493 140603635533376 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 18:02:15.468674 140603635533376 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.587171s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9/items I0422 18:02:15.469367 140603635533376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9/items I0422 18:02:15.716359 140603635533376 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9 I0422 18:02:16.442121 140603635533376 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9`. I0422 18:02:16.442857 140603635533376 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.491505s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_06_grad_accum/checkpoints/9 I0422 18:02:16.442931 140603635533376 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.680082s. I0422 18:02:16.443178 140602417088064 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 18:02:16.443303 140602417088064 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 18:02:16.443364 140602417088064 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 18:02:16.443405 140602417088064 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 18:02:16.443539 140634764187456 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. I0422 18:02:16.443675 140634764187456 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 18:02:16.444550 140634764187456 metric_logger.py:196] completed step: 9, seconds: 0.424, TFLOP/s/device: 128.183, Tokens/s/device: 19321.210, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3551.987 Per train step: Total TFLOPs: 54.35 split as 93.93% learnable weight flops and 6.07% attention flops