2026-04-21 15:03:15.618549: 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 00:00:1776783795.628934 1490705 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776783795.632211 1490705 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776783795.641806 1490705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783795.641817 1490705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783795.641819 1490705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776783795.641821 1490705 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 15:03:19.251542: 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) I0421 15:03:19.368283 140233011838080 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:03:47.000226 140233011838080 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:03:47.000343 140233011838080 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:03:47.000379 140233011838080 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 I0421 15:03:47.000404 140233011838080 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:03:47.000426 140233011838080 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 15:03:47.000503 140233011838080 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:03:47.358263 140233011838080 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:03:47.358643 140233011838080 checkpointing.py:677] Setting up checkpoint logger... I0421 15:03:47.358695 140233011838080 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:03:47.358815 140233011838080 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:03:47.359899 140233011838080 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 0x7f8a435a3230>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:03:49.726640 140233011838080 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:03:49.727022 140233011838080 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 0x7f889632ebd0>}, handler_registry=None I0421 15:03:49.727515 140233011838080 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f889632ebd0>` 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`. I0421 15:03:49.727560 140233011838080 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f8895f727b0>` 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`. I0421 15:03:49.727591 140233011838080 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 0x7f889632ebd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f889632ebd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f8895f727b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f8895f727b0>}). I0421 15:03:49.727925 140233011838080 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:03:49.728002 140233011838080 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7f8895f95260> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 15:03:49.837777 140233011838080 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints I0421 15:03:49.838096 140233011838080 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_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f889632c890> I0421 15:03:49.838197 140233011838080 checkpointing.py:301] Checkpoint manager created! I0421 15:03:50.975580 140233011838080 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:03:50.975680 140233011838080 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:03:51.257090 140233011838080 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:03:51.257175 140233011838080 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:03:51.269387 140233011838080 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:03:51.269442 140233011838080 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:03:51.288399 140233011838080 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:03:51.288460 140233011838080 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:03:51.300610 140233011838080 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:03:51.300660 140233011838080 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:03:51.312787 140233011838080 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 15:03:51.312842 140233011838080 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:03:51.332058 140233011838080 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:03:51.332116 140233011838080 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:03:51.347351 140233011838080 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:03:51.347402 140233011838080 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:03:51.497271 140233011838080 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:03:51.497366 140233011838080 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 I0421 15:03:52.092533 140233011838080 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:03:52.092645 140233011838080 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:03:52.092687 140233011838080 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 15:03:52.092732 140233011838080 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 15:03:52.092772 140233011838080 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:03:52.092806 140233011838080 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0421 15:03:52.092846 140233011838080 maxtext_utils.py:1654] 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) I0421 15:03:52.092885 140233011838080 maxtext_utils.py:1654] 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') I0421 15:03:52.092913 140233011838080 maxtext_utils.py:1654] 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) I0421 15:03:52.092939 140233011838080 maxtext_utils.py:1654] 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) I0421 15:03:52.092972 140233011838080 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:03:52.093006 140233011838080 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:03:52.109658 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0421 15:03:52.109710 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0421 15:03:52.120988 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0421 15:03:52.121038 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0421 15:03:52.142759 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0421 15:03:52.142822 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0421 15:03:52.153723 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0421 15:03:52.153769 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0421 15:03:52.175528 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0421 15:03:52.175585 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0421 15:03:52.186522 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0421 15:03:52.186567 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0421 15:03:52.219448 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0421 15:03:52.219513 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0421 15:03:52.230429 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0421 15:03:52.230475 140233011838080 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0421 15:03:52.762110 140233011838080 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:03:52.762198 140233011838080 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:03:52.773941 140233011838080 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:03:52.773990 140233011838080 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:04:00.937554 140233011838080 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. I0421 15:04:00.938621 140233011838080 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:04:00.938969 140233011838080 metric_logger.py:301] number parameters: 1.104 billion I0421 15:04:03.773250 140233011838080 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0421 15:04:04.191686 140233011838080 checkpointing.py:776] Waited 0.4184126853942871 seconds for step 0 to finish before starting checkpointing. I0421 15:04:04.192212 140233011838080 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:04:04.192399 140233011838080 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0421 15:04:04.192803 140233011838080 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0. I0421 15:04:04.277369 140233011838080 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0421 15:04:04.369429 140148761167424 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0 I0421 15:04:04.376947 140233011838080 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:04:04.377060 140233011838080 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 I0421 15:04:05.051258 140233011838080 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.674793s I0421 15:04:05.051823 140233011838080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 15.959 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 773 milliseconds) (per-host) I0421 15:04:05.051882 140233011838080 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.773410s (batch_requests_ready=0.094430s, total_serialization_initiated=0.678558s, others=0.000422s) I0421 15:04:05.051954 140233011838080 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.774021s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.774001) I0421 15:04:05.052840 140148601783872 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:04:05.052974 140233011838080 async_checkpointer.py:561] Finished blocking save. Time taken: 0.860530s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0. I0421 15:04:05.053227 140233011838080 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:04:05.053479 140148771653184 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:04:05.053624 140233011838080 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776783844.192198, 'wait_for_prev_duration_secs': 4.4345855712890625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776783844.1924214, 'checkpointer_blocking_duration_secs': 0.8607077598571777, 'get_old_steps_start_time': 1776783845.0531535, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776783844.1921144, 'checkpoint_manager_blocking_duration_secs': 0.8614671230316162} I0421 15:04:05.053780 140233011838080 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0421 15:04:05.053832 140233011838080 max_utils.py:750] Memstats: After params initialized: I0421 15:04:05.053942 140233011838080 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:04:05.053973 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:04:05.053994 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:04:05.054013 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:04:05.054031 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:04:05.054065 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:04:05.054084 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:04:05.054102 140233011838080 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:04:05.167618 140148750681664 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/items I0421 15:04:05.317188 140148591298112 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783845064354853, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776783845064354853, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0421 15:04:05.468078 140233011838080 metric_logger.py:196] completed step: 0, seconds: 2.833, TFLOP/s/device: 19.181, Tokens/s/device: 2891.202, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0421 15:04:05.469072 140233011838080 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/tensorboard/' I0421 15:04:05.868164 1493005 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:04:05.868254 1493005 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:04:05.894428 140233011838080 metric_logger.py:196] completed step: 1, seconds: 1.692, TFLOP/s/device: 32.117, Tokens/s/device: 4841.050, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0421 15:04:06.324415 140233011838080 metric_logger.py:196] completed step: 2, seconds: 0.008, TFLOP/s/device: 7000.968, Tokens/s/device: 1055262.141, total_weights: 65536, loss: 10.261, lm_loss: 10.261, perplexity: 28586.895 I0421 15:04:06.594783 140148715030080 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/items/array_metadatas/process_0 I0421 15:04:06.748851 140233011838080 metric_logger.py:196] completed step: 3, seconds: 0.427, TFLOP/s/device: 127.263, Tokens/s/device: 19182.451, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16845.295 I0421 15:04:34.223836 140233011838080 metric_logger.py:196] completed step: 4, seconds: 0.430, TFLOP/s/device: 126.318, Tokens/s/device: 19039.960, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10654.189 I0421 15:04:34.232955 140233011838080 metric_logger.py:196] completed step: 5, seconds: 0.424, TFLOP/s/device: 128.079, Tokens/s/device: 19305.502, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7241.146 I0421 15:04:34.655484 140233011838080 metric_logger.py:196] completed step: 6, seconds: 27.477, TFLOP/s/device: 1.978, Tokens/s/device: 298.143, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5368.870 I0421 15:04:35.079699 140233011838080 metric_logger.py:196] completed step: 7, seconds: 0.005, TFLOP/s/device: 10485.919, Tokens/s/device: 1580551.804, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4361.125 I0421 15:04:35.498661 140233011838080 metric_logger.py:196] completed step: 8, seconds: 0.423, TFLOP/s/device: 128.386, Tokens/s/device: 19351.745, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3832.059 I0421 15:04:35.927235 140233011838080 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0421 15:04:35.928080 140233011838080 checkpointing.py:776] Waited 0.0008435249328613281 seconds for step 9 to finish before starting checkpointing. I0421 15:04:35.928379 140233011838080 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:04:43.261222 140148704544320 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.390957s (commit=36.971430s, array_metadata_write=0.419528s) I0421 15:04:43.262431 140148601783872 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 324.154 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 38 seconds) (per-host) I0421 15:04:43.262517 140148601783872 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.209482s. I0421 15:04:43.517406 140148601783872 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783845064354853, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0421 15:04:43.704796 140148601783872 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:04:43.910094 140148591298112 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783845064354853, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0421 15:04:44.156461 140148601783872 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:04:44.157153 140148601783872 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.597795s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/items I0421 15:04:44.157945 140148601783872 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0/items I0421 15:04:44.385916 140148601783872 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0 I0421 15:04:45.063634 140148601783872 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0`. I0421 15:04:45.064356 140148601783872 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 40.871922s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/0 I0421 15:04:45.064435 140148601783872 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.011405s. I0421 15:04:45.064628 140148771653184 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:04:45.064747 140148771653184 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:04:45.064808 140148771653184 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:04:45.064856 140148771653184 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 15:04:45.065016 140233011838080 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. W0421 15:04:45.065182 140233011838080 checkpoint_manager.py:1441] Waiting for previous save to complete took 9.136818 seconds. If this number is high, consider checkpointing less frequently. I0421 15:04:45.066087 140233011838080 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0421 15:04:45.066419 140233011838080 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9. I0421 15:04:45.236639 140148771653184 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9 I0421 15:04:45.250315 140233011838080 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:04:45.250420 140233011838080 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 I0421 15:04:45.962203 140148704544320 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9/items I0421 15:04:55.580914 140233011838080 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 10.331110s I0421 15:04:55.585402 140233011838080 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.182 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 10 seconds) (per-host) I0421 15:04:55.585466 140233011838080 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 10.437704s (batch_requests_ready=0.098971s, total_serialization_initiated=10.334404s, others=0.004329s) I0421 15:04:55.585541 140233011838080 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 10.438295s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=10.438276) I0421 15:04:55.587403 140148771653184 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0421 15:04:55.587513 140233011838080 async_checkpointer.py:561] Finished blocking save. Time taken: 10.521380s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9. I0421 15:04:55.587717 140233011838080 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 15:04:55.587955 140148601783872 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0421 15:04:55.588093 140233011838080 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776783875.9283268, 'wait_for_prev_duration_secs': 9.136817693710327, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776783885.0661128, 'checkpointer_blocking_duration_secs': 10.52151608467102, 'get_old_steps_start_time': 1776783895.5876436, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776783875.9282765, 'checkpoint_manager_blocking_duration_secs': 19.659790754318237} I0421 15:04:55.588210 140233011838080 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0421 15:04:55.588243 140233011838080 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 15:04:56.308171 140146846467648 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9/items/array_metadatas/process_0 I0421 15:05:29.226227 140148761167424 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.639977s (commit=33.193479s, array_metadata_write=0.446498s) I0421 15:05:29.227512 140148771653184 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 286.680 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0421 15:05:29.227627 140148771653184 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.640067s. I0421 15:05:29.648574 140148771653184 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:05:30.069477 140148771653184 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:05:30.069660 140148771653184 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.563107s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9/items I0421 15:05:30.070378 140148771653184 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9/items I0421 15:05:30.304563 140148771653184 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9 I0421 15:05:30.974539 140148771653184 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9`. I0421 15:05:30.975291 140148771653184 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 45.909163s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_06_grad_accum/checkpoints/9 I0421 15:05:30.975368 140148771653184 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 35.387810s. I0421 15:05:30.975601 140148601783872 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 15:05:30.975731 140148601783872 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0421 15:05:30.975799 140148601783872 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:05:30.975841 140148601783872 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 15:05:30.976300 140233011838080 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. I0421 15:05:30.976937 140233011838080 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:05:30.978100 140233011838080 metric_logger.py:196] completed step: 9, seconds: 0.424, TFLOP/s/device: 128.048, Tokens/s/device: 19300.726, 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