2026-04-21 15:40:26.656417: 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 00:00:1776786026.666437 1530271 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776786026.669508 1530271 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776786026.679447 1530271 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786026.679458 1530271 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786026.679459 1530271 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776786026.679461 1530271 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-21 15:40:30.190433: 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 15:40:30.309480 139692938898560 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0421 15:40:58.236733 139692938898560 max_utils.py:800] System Information: Jax Version: 0.8.1 I0421 15:40:58.236913 139692938898560 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0421 15:40:58.236950 139692938898560 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 15:40:58.236981 139692938898560 train_utils.py:335] WARNING: 'dataset_path' might be pointing your local file system I0421 15:40:58.237003 139692938898560 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 15:40:58.237133 139692938898560 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0421 15:40:58.592848 139692938898560 maxtext_utils.py:1551] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 15:40:58.593275 139692938898560 checkpointing.py:677] Setting up checkpoint logger... I0421 15:40:58.593329 139692938898560 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 15:40:58.593402 139692938898560 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 15:40:58.594064 139692938898560 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 0x7f0c7fba1df0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 15:41:00.256888 139692938898560 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 15:41:00.257250 139692938898560 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 0x7f0ad9f0ea20>}, handler_registry=None I0421 15:41:00.257634 139692938898560 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f0ad9f0ea20>` 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 15:41:00.257678 139692938898560 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f0ad9b82f60>` 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 15:41:00.257709 139692938898560 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 0x7f0ad9f0ea20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f0ad9f0ea20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f0ad9b82f60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f0ad9b82f60>}). I0421 15:41:00.258020 139692938898560 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0421 15:41:00.668120 139609319147072 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776785875023187510, 'commit_timestamp_nsecs': 1776785916078598517, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/0/_CHECKPOINT_METADATA I0421 15:41:00.749563 139609319147072 checkpoint_manager.py:1847] Missing metrics for step 0 I0421 15:41:00.823811 139692938898560 checkpoint_manager.py:1788] Found 3 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints I0421 15:41:00.824435 139692938898560 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=False, 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=5), 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_14_async_ckpt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f0ad9b82b70> I0421 15:41:00.824536 139692938898560 checkpointing.py:301] Checkpoint manager created! I0421 15:41:01.472314 139692938898560 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 15:41:01.472404 139692938898560 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:41:01.749997 139692938898560 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:41:01.750093 139692938898560 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:41:01.762065 139692938898560 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 15:41:01.762113 139692938898560 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0421 15:41:01.780851 139692938898560 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 15:41:01.780907 139692938898560 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:41:01.792934 139692938898560 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 15:41:01.792989 139692938898560 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:41:01.804961 139692938898560 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 15:41:01.805014 139692938898560 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:41:01.824013 139692938898560 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 15:41:01.824096 139692938898560 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0421 15:41:01.839296 139692938898560 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 15:41:01.839347 139692938898560 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0421 15:41:01.988054 139692938898560 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 15:41:01.988148 139692938898560 checkpointing.py:581] restoring from this run's directory step 9 I0421 15:41:02.164676 139692938898560 checkpointer.py:304] Restoring checkpoint from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/9. I0421 15:41:02.680829 1531732 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0421 15:41:02.680908 1531732 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0421 15:41:09.072461 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/read/gbytes_per_sec: 1.879 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 6 seconds) (per-host) I0421 15:41:09.072800 139692938898560 checkpointer.py:318] Finished restoring checkpoint in 7.00 seconds from gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/9. I0421 15:41:09.072896 139692938898560 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints', 'checkpointer_start_time': 1776786062.0746233, 'checkpointer_duration_secs': 6.998222589492798, 'checkpoint_manager_start_time': 1776786061.9883518, 'checkpoint_manager_duration_secs': 7.084497928619385} [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 15:41:09.496799 139692938898560 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0421 15:41:09.496920 139692938898560 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 15:41:09.496962 139692938898560 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 15:41:09.497007 139692938898560 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 15:41:09.497069 139692938898560 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 15:41:09.497110 139692938898560 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 15:41:09.497151 139692938898560 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 15:41:09.497190 139692938898560 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 15:41:09.497219 139692938898560 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 15:41:09.497247 139692938898560 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 15:41:09.497281 139692938898560 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 15:41:09.497314 139692938898560 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 15:41:10.007245 139692938898560 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:41:10.007335 139692938898560 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:41:10.018628 139692938898560 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0421 15:41:10.018676 139692938898560 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0421 15:41:10.674884 139692938898560 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0421 15:41:10.675495 139692938898560 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0421 15:41:10.675665 139692938898560 metric_logger.py:301] number parameters: 1.104 billion I0421 15:41:12.874220 139692938898560 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 15:41:12.989750 139692938898560 checkpointing.py:776] Waited 0.11551141738891602 seconds for step 10 to finish before starting checkpointing. I0421 15:41:12.990245 139692938898560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:41:12.990417 139692938898560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0421 15:41:12.990482 139692938898560 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10. I0421 15:41:13.173988 139692938898560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10 I0421 15:41:13.993750 139692938898560 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776786073751928670, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776786073751928670, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10/_CHECKPOINT_METADATA I0421 15:41:13.994870 139692938898560 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient W0421 15:41:13.995258 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:41:14.081489 139608635475520 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10/items I0421 15:41:14.633404 139692938898560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:41:14.633518 139692938898560 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 15:41:23.312504 139692938898560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.679534s I0421 15:41:23.317101 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.408 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host) I0421 15:41:23.317171 139692938898560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.767465s (batch_requests_ready=0.080461s, total_serialization_initiated=8.682544s, others=0.004459s) I0421 15:41:23.317258 139692938898560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.322462s (all_items=0.000035s, per_item={'items': '0.00003505'}, temp_paths=9.322427) W0421 15:41:23.317300 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:41:24.076139 139608645961280 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_14_async_ckpt_false/checkpoints/10/items/array_metadatas/process_0 I0421 15:42:08.636962 139608624989760 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 45.319265s (commit=44.833592s, array_metadata_write=0.485673s) I0421 15:42:08.638458 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 233.631 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 54 seconds) (per-host) I0421 15:42:09.264706 139692938898560 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776786073751928670, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10/_CHECKPOINT_METADATA I0421 15:42:09.413412 139692938898560 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:42:09.873836 139692938898560 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:42:09.874018 139692938898560 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.608819s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10/items I0421 15:42:09.874860 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10/items I0421 15:42:10.140105 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10 I0421 15:42:10.847057 139692938898560 atomicity.py:794] [process=0][thread=MainThread] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10`. I0421 15:42:10.847833 139692938898560 checkpointer.py:290] Finished synchronous save in 57.86 seconds to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/10 I0421 15:42:10.848035 139692938898560 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:42:10.848098 139692938898560 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 15:42:10.848123 139692938898560 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=10] Finished synchronous save. I0421 15:42:10.848163 139692938898560 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776786072.990232, 'wait_for_prev_duration_secs': 4.029273986816406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776786072.990439, 'checkpointer_blocking_duration_secs': 57.857444047927856, 'get_old_steps_start_time': 1776786130.8479145, 'get_old_steps_duration_secs': 3.7670135498046875e-05, 'checkpoint_manager_blocking_start_time': 1776786072.9901423, 'checkpoint_manager_blocking_duration_secs': 57.857996702194214} I0421 15:42:10.848281 139692938898560 checkpointing.py:410] Saved a checkpoint at step 10. I0421 15:42:10.848333 139692938898560 max_utils.py:750] Memstats: After params initialized: I0421 15:42:10.848391 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0421 15:42:10.848417 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0421 15:42:10.848437 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0421 15:42:10.848456 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0421 15:42:10.848473 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0421 15:42:10.848490 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0421 15:42:10.848508 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0421 15:42:10.848525 139692938898560 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0421 15:42:11.117019 139692938898560 metric_logger.py:196] completed step: 10, seconds: 2.198, TFLOP/s/device: 6.182, Tokens/s/device: 931.778, total_weights: 16384, loss: 3.602, lm_loss: 3.602, perplexity: 36.686 I0421 15:42:11.231509 139692938898560 metric_logger.py:196] completed step: 11, seconds: 58.238, TFLOP/s/device: 0.233, Tokens/s/device: 35.166, total_weights: 16384, loss: 3.101, lm_loss: 3.101, perplexity: 22.211 I0421 15:42:11.350401 139692938898560 metric_logger.py:196] completed step: 12, seconds: 0.008, TFLOP/s/device: 1653.740, Tokens/s/device: 249269.718, total_weights: 16384, loss: 2.684, lm_loss: 2.684, perplexity: 14.644 I0421 15:42:11.469361 139692938898560 metric_logger.py:196] completed step: 13, seconds: 0.115, TFLOP/s/device: 118.066, Tokens/s/device: 17796.161, total_weights: 16384, loss: 2.348, lm_loss: 2.348, perplexity: 10.461 I0421 15:42:11.473960 139692938898560 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 15:42:11.706488 139692938898560 checkpointing.py:776] Waited 0.23248815536499023 seconds for step 15 to finish before starting checkpointing. I0421 15:42:11.706905 139692938898560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:42:11.707081 139692938898560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 15 I0421 15:42:11.707144 139692938898560 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15. I0421 15:42:11.883278 139692938898560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15 W0421 15:42:12.737981 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:42:12.824152 139608624989760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15/items I0421 15:42:13.378870 139692938898560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:42:13.378986 139692938898560 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 15:42:14.536575 139692938898560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.158088s I0421 15:42:14.541244 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 9.876 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0421 15:42:14.541308 139692938898560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.249716s (batch_requests_ready=0.083930s, total_serialization_initiated=1.161270s, others=0.004516s) I0421 15:42:14.541407 139692938898560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.803796s (all_items=0.000042s, per_item={'items': '0.00004196'}, temp_paths=1.803754) W0421 15:42:14.541446 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:42:15.293036 139608635475520 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_14_async_ckpt_false/checkpoints/15/items/array_metadatas/process_0 I0421 15:42:50.913485 139608645961280 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.371803s (commit=35.905014s, array_metadata_write=0.466789s) I0421 15:42:50.914790 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 335.878 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0421 15:42:51.712318 139692938898560 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:42:52.142809 139692938898560 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:42:52.143121 139692938898560 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.571066s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15/items I0421 15:42:52.143858 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15/items I0421 15:42:52.381384 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15 I0421 15:42:53.077019 139692938898560 atomicity.py:794] [process=0][thread=MainThread] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15`. I0421 15:42:53.077802 139692938898560 checkpointer.py:290] Finished synchronous save in 41.37 seconds to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/15 I0421 15:42:53.077977 139692938898560 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:42:53.078020 139692938898560 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 15:42:53.078060 139692938898560 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=15] Finished synchronous save. I0421 15:42:53.078097 139692938898560 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776786131.7068794, 'wait_for_prev_duration_secs': 5.14984130859375e-05, 'time_between_consecutive_saves_sec': 0.8587648868560791, 'checkpointer_blocking_start_time': 1776786131.707103, 'checkpointer_blocking_duration_secs': 41.37074899673462, 'get_old_steps_start_time': 1776786173.0778797, 'get_old_steps_duration_secs': 3.3855438232421875e-05, 'checkpoint_manager_blocking_start_time': 1776786131.7068114, 'checkpoint_manager_blocking_duration_secs': 41.37126326560974} I0421 15:42:53.078214 139692938898560 checkpointing.py:410] Saved a checkpoint at step 15. I0421 15:42:53.082656 139692938898560 metric_logger.py:196] completed step: 14, seconds: 0.119, TFLOP/s/device: 114.324, Tokens/s/device: 17232.095, total_weights: 16384, loss: 2.083, lm_loss: 2.083, perplexity: 8.030 I0421 15:42:53.093455 139692938898560 metric_logger.py:196] completed step: 15, seconds: 0.119, TFLOP/s/device: 114.190, Tokens/s/device: 17211.964, total_weights: 16384, loss: 1.880, lm_loss: 1.880, perplexity: 6.551 I0421 15:42:53.207693 139692938898560 metric_logger.py:196] completed step: 16, seconds: 41.615, TFLOP/s/device: 0.326, Tokens/s/device: 49.213, total_weights: 16384, loss: 1.726, lm_loss: 1.726, perplexity: 5.616 I0421 15:42:53.326686 139692938898560 metric_logger.py:196] completed step: 17, seconds: 0.008, TFLOP/s/device: 1635.230, Tokens/s/device: 246479.721, total_weights: 16384, loss: 1.610, lm_loss: 1.610, perplexity: 5.002 I0421 15:42:53.445830 139692938898560 metric_logger.py:196] completed step: 18, seconds: 0.116, TFLOP/s/device: 117.617, Tokens/s/device: 17728.532, total_weights: 16384, loss: 1.521, lm_loss: 1.521, perplexity: 4.577 I0421 15:42:53.563638 139692938898560 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 15:42:53.564405 139692938898560 checkpointing.py:776] Waited 0.0007898807525634766 seconds for step 19 to finish before starting checkpointing. I0421 15:42:53.564648 139692938898560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:42:53.564794 139692938898560 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0421 15:42:53.564852 139692938898560 checkpointer.py:241] [process=0] Started saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19. I0421 15:42:53.743607 139692938898560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19 W0421 15:42:54.581413 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:42:54.672664 139608624989760 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19/items I0421 15:42:55.214092 139692938898560 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0421 15:42:55.214209 139692938898560 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 15:42:56.098180 139692938898560 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.884488s I0421 15:42:56.098600 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.672 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 973 milliseconds) (per-host) I0421 15:42:56.098724 139692938898560 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.974010s (batch_requests_ready=0.085958s, total_serialization_initiated=0.887727s, others=0.000326s) I0421 15:42:56.098796 139692938898560 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.517779s (all_items=0.000039s, per_item={'items': '0.00003886'}, temp_paths=1.517740) W0421 15:42:56.098832 139692938898560 future.py:328] [process=0][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0421 15:42:56.861166 139608645961280 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_14_async_ckpt_false/checkpoints/19/items/array_metadatas/process_0 I0421 15:43:26.089843 139609262523968 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.986517s (commit=29.544425s, array_metadata_write=0.442091s) I0421 15:43:26.091163 139692938898560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 408.082 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 30 seconds) (per-host) I0421 15:43:26.872193 139692938898560 array_metadata_store.py:367] [process=0][thread=MainThread] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0421 15:43:27.308615 139692938898560 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0421 15:43:27.308932 139692938898560 base_pytree_checkpoint_handler.py:1342] [process=0][thread=MainThread] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.585197s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19/items I0421 15:43:27.309773 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19/items I0421 15:43:27.547454 139692938898560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19 I0421 15:43:28.219734 139692938898560 atomicity.py:794] [process=0][thread=MainThread] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19`. I0421 15:43:28.220569 139692938898560 checkpointer.py:290] Finished synchronous save in 34.66 seconds to gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints/19 I0421 15:43:28.220759 139692938898560 checkpoint_manager.py:2103] [process=0][thread=MainThread][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 15:43:28.220812 139692938898560 checkpoint_manager.py:2112] [process=0][thread=MainThread][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 15:43:28.220836 139692938898560 checkpoint_manager.py:1570] [process=0][thread=MainThread][step=19] Finished synchronous save. I0421 15:43:28.220886 139692938898560 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260421_144657/linen_main_20260421_144657_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1776786173.5646276, 'wait_for_prev_duration_secs': 4.506111145019531e-05, 'time_between_consecutive_saves_sec': 0.48659205436706543, 'checkpointer_blocking_start_time': 1776786173.564814, 'checkpointer_blocking_duration_secs': 34.6558051109314, 'get_old_steps_start_time': 1776786208.2206504, 'get_old_steps_duration_secs': 2.193450927734375e-05, 'checkpoint_manager_blocking_start_time': 1776786173.564596, 'checkpoint_manager_blocking_duration_secs': 34.656264305114746} I0421 15:43:28.220993 139692938898560 checkpointing.py:410] Saved a checkpoint at step 19. I0421 15:43:28.221021 139692938898560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:43:28.221041 139692938898560 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 15:43:28.221986 139692938898560 metric_logger.py:196] completed step: 19, seconds: 0.119, TFLOP/s/device: 114.309, Tokens/s/device: 17229.920, total_weights: 16384, loss: 1.449, lm_loss: 1.449, perplexity: 4.260 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops