2026-04-23 01:35:25.569945: 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 00:00:1776908125.583224 2272621 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776908125.587027 2272621 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776908125.598544 2272621 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908125.598560 2272621 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908125.598562 2272621 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776908125.598563 2272621 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'} 2026-04-23 01:35:29.256392: 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) I0423 01:35:29.392316 123767550224192 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 01:36:15.245840 123767550224192 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 01:36:15.246143 123767550224192 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 01:36:15.246197 123767550224192 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 I0423 01:36:15.246231 123767550224192 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 01:36:15.246253 123767550224192 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 01:36:15.246431 123767550224192 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 01:36:15.904500 123767550224192 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 01:36:15.904778 123767550224192 checkpointing.py:677] Setting up checkpoint logger... I0423 01:36:15.904823 123767550224192 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 01:36:15.904860 123767550224192 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 01:36:15.905306 123767550224192 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 0x70909bdaf050>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 01:36:18.363359 123767550224192 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 01:36:18.363906 123767550224192 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 0x708ec1b3d970>}, handler_registry=None I0423 01:36:18.364285 123767550224192 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x708ec1b3d970>` 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`. I0423 01:36:18.364334 123767550224192 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x708f3bd8d880>` 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`. I0423 01:36:18.364366 123767550224192 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 0x708ec1b3d970>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x708ec1b3d970>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x708f3bd8d880>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x708f3bd8d880>}). I0423 01:36:18.364657 123767550224192 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 01:36:18.364714 123767550224192 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x708ec1b54900> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 01:36:18.492796 123767550224192 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints I0423 01:36:18.493127 123767550224192 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_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x708f3e50f6b0> I0423 01:36:18.493446 123767550224192 checkpointing.py:301] Checkpoint manager created! I0423 01:36:19.400036 123767550224192 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 01:36:19.400153 123767550224192 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:36:19.682996 123767550224192 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:36:19.683099 123767550224192 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:36:19.695311 123767550224192 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 01:36:19.695360 123767550224192 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 01:36:19.714428 123767550224192 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:36:19.714496 123767550224192 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:36:19.726702 123767550224192 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:36:19.726760 123767550224192 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:36:19.738867 123767550224192 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 01:36:19.738920 123767550224192 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:36:19.758336 123767550224192 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 01:36:19.758395 123767550224192 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 01:36:19.773788 123767550224192 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 01:36:19.773841 123767550224192 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 01:36:19.924299 123767550224192 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 01:36:19.924402 123767550224192 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 I0423 01:36:20.547587 123767550224192 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 01:36:20.547704 123767550224192 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) I0423 01:36:20.547756 123767550224192 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) I0423 01:36:20.547805 123767550224192 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') I0423 01:36:20.547847 123767550224192 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) I0423 01:36:20.547880 123767550224192 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) I0423 01:36:20.547922 123767550224192 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) I0423 01:36:20.547963 123767550224192 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') I0423 01:36:20.547993 123767550224192 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) I0423 01:36:20.548019 123767550224192 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) I0423 01:36:20.548069 123767550224192 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 01:36:20.548112 123767550224192 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 01:36:20.564419 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0423 01:36:20.564478 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0423 01:36:20.575499 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0423 01:36:20.575548 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0423 01:36:20.597534 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0423 01:36:20.597600 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0423 01:36:20.608555 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0423 01:36:20.608611 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0423 01:36:20.630554 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0423 01:36:20.630608 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0423 01:36:20.641600 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0423 01:36:20.641648 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0423 01:36:20.674410 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0423 01:36:20.674475 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0423 01:36:20.685484 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0423 01:36:20.685534 123767550224192 gradient_accumulation.py:68] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0423 01:36:21.216567 123767550224192 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:36:21.216657 123767550224192 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:36:21.228576 123767550224192 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 01:36:21.228626 123767550224192 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 01:36:22.248321 123767550224192 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. I0423 01:36:22.248908 123767550224192 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 01:36:22.249081 123767550224192 metric_logger.py:301] number parameters: 1.104 billion I0423 01:36:25.399228 123767550224192 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 01:36:25.816742 123767550224192 checkpointing.py:776] Waited 0.41749000549316406 seconds for step 0 to finish before starting checkpointing. I0423 01:36:25.817387 123767550224192 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:36:25.817537 123767550224192 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 01:36:25.817946 123767550224192 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0. I0423 01:36:25.900594 123767550224192 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 01:36:25.984898 123767550224192 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:36:25.985001 123767550224192 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 I0423 01:36:25.991475 123736342660672 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0 I0423 01:36:26.788393 123736193762880 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/items I0423 01:36:26.928770 123735956784704 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908186679182973, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776908186679182973, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0423 01:36:27.126055 123767550224192 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 1.141666s I0423 01:36:27.126546 123767550224192 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 10.075 GiB/s (total gbytes: 12.3 GiB) (time elapsed: a second) (per-host) I0423 01:36:27.126598 123767550224192 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 1.224946s (batch_requests_ready=0.079324s, total_serialization_initiated=1.145249s, others=0.000373s) I0423 01:36:27.126657 123767550224192 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.225512s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=1.225492) I0423 01:36:27.127613 123735736583744 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:36:27.127720 123767550224192 async_checkpointer.py:561] Finished blocking save. Time taken: 1.310143s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0. I0423 01:36:27.127907 123767550224192 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:36:27.128118 123736353146432 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:36:27.128227 123767550224192 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776908185.8173733, 'wait_for_prev_duration_secs': 4.2438507080078125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776908185.8175583, 'checkpointer_blocking_duration_secs': 1.3102734088897705, 'get_old_steps_start_time': 1776908187.1278465, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776908185.8172903, 'checkpoint_manager_blocking_duration_secs': 1.3109116554260254} I0423 01:36:27.128334 123767550224192 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 01:36:27.128375 123767550224192 max_utils.py:750] Memstats: After params initialized: I0423 01:36:27.128459 123767550224192 max_utils.py:756] Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0)) I0423 01:36:27.128488 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0423 01:36:27.128509 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0423 01:36:27.128528 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0423 01:36:27.128546 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0423 01:36:27.128564 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0423 01:36:27.128583 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0423 01:36:27.128600 123767550224192 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0423 01:36:27.256330 2274758 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 01:36:27.256383 2274758 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 01:36:27.483337 123767550224192 metric_logger.py:196] completed step: 0, seconds: 3.150, TFLOP/s/device: 17.256, Tokens/s/device: 2600.953, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0423 01:36:27.484371 123767550224192 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/tensorboard/' I0423 01:36:27.906844 123767550224192 metric_logger.py:196] completed step: 1, seconds: 2.082, TFLOP/s/device: 26.107, Tokens/s/device: 3935.109, total_weights: 65536, loss: 10.871, lm_loss: 10.871, perplexity: 52637.648 I0423 01:36:28.252509 123735946298944 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/items/array_metadatas/process_0 I0423 01:36:28.338940 123767550224192 metric_logger.py:196] completed step: 2, seconds: 0.008, TFLOP/s/device: 7175.669, Tokens/s/device: 1081594.930, total_weights: 65536, loss: 10.261, lm_loss: 10.261, perplexity: 28586.895 I0423 01:36:28.761538 123767550224192 metric_logger.py:196] completed step: 3, seconds: 0.425, TFLOP/s/device: 127.747, Tokens/s/device: 19255.404, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16845.295 I0423 01:37:03.985286 123767550224192 metric_logger.py:196] completed step: 4, seconds: 0.431, TFLOP/s/device: 126.087, Tokens/s/device: 19005.241, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10654.189 I0423 01:37:03.995154 123767550224192 metric_logger.py:196] completed step: 5, seconds: 0.424, TFLOP/s/device: 128.196, Tokens/s/device: 19323.079, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7241.146 I0423 01:37:04.417381 123767550224192 metric_logger.py:196] completed step: 6, seconds: 35.225, TFLOP/s/device: 1.543, Tokens/s/device: 232.564, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5368.870 I0423 01:37:04.838716 123767550224192 metric_logger.py:196] completed step: 7, seconds: 0.006, TFLOP/s/device: 9811.973, Tokens/s/device: 1478967.323, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4361.125 I0423 01:37:05.262551 123767550224192 metric_logger.py:196] completed step: 8, seconds: 0.423, TFLOP/s/device: 128.382, Tokens/s/device: 19351.196, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3832.059 I0423 01:37:05.691491 123767550224192 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 01:37:05.692692 123767550224192 checkpointing.py:776] Waited 0.0012366771697998047 seconds for step 9 to finish before starting checkpointing. I0423 01:37:05.693060 123767550224192 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:37:12.911681 123736183277120 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 45.653393s (commit=45.166184s, array_metadata_write=0.487209s) I0423 01:37:12.912926 123735736583744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 268.804 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host) I0423 01:37:12.913062 123735736583744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.785275s. I0423 01:37:13.162527 123735736583744 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908186679182973, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0423 01:37:13.365272 123735736583744 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 01:37:13.581206 123735956784704 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776908186679182973, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA I0423 01:37:13.783240 123735736583744 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:37:13.783424 123735736583744 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.585971s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/items I0423 01:37:13.784210 123735736583744 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0/items I0423 01:37:14.017779 123735736583744 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0 I0423 01:37:14.716335 123735736583744 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0`. I0423 01:37:14.717110 123735736583744 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.899537s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/0 I0423 01:37:14.717200 123735736583744 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 47.589433s. I0423 01:37:14.717440 123736353146432 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:37:14.717565 123736353146432 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:37:14.717652 123736353146432 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:37:14.717705 123736353146432 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 01:37:14.725039 123767550224192 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. W0423 01:37:14.725143 123767550224192 checkpoint_manager.py:1441] Waiting for previous save to complete took 9.032146 seconds. If this number is high, consider checkpointing less frequently. I0423 01:37:14.726414 123767550224192 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 01:37:14.726741 123767550224192 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9. I0423 01:37:14.897468 123767550224192 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0423 01:37:14.897624 123735736583744 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9 I0423 01:37:14.897714 123767550224192 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 I0423 01:37:15.586295 123735946298944 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9/items I0423 01:37:20.467573 123767550224192 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 5.570774s I0423 01:37:20.472509 123767550224192 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.181 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 5 seconds) (per-host) I0423 01:37:20.472579 123767550224192 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 5.658677s (batch_requests_ready=0.079794s, total_serialization_initiated=5.574064s, others=0.004819s) I0423 01:37:20.472657 123767550224192 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 5.659552s (all_items=0.000029s, per_item={'items': '0.00002861'}, temp_paths=5.659523) I0423 01:37:20.475253 123736353146432 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 01:37:20.475366 123767550224192 async_checkpointer.py:561] Finished blocking save. Time taken: 5.748905s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9. I0423 01:37:20.475589 123767550224192 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 01:37:20.475804 123736183277120 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 01:37:20.475914 123767550224192 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776908225.6929693, 'wait_for_prev_duration_secs': 9.032145738601685, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776908234.7264383, 'checkpointer_blocking_duration_secs': 5.749049186706543, 'get_old_steps_start_time': 1776908240.4755054, 'get_old_steps_duration_secs': 4.172325134277344e-05, 'checkpoint_manager_blocking_start_time': 1776908225.6929138, 'checkpoint_manager_blocking_duration_secs': 14.782975435256958} I0423 01:37:20.476040 123767550224192 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 01:37:20.476090 123767550224192 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 01:37:21.215660 123736162305600 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9/items/array_metadatas/process_0 I0423 01:37:56.387639 123736151819840 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.914252s (commit=35.450095s, array_metadata_write=0.464157s) I0423 01:37:56.388867 123736353146432 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 303.952 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 41 seconds) (per-host) I0423 01:37:56.388983 123736353146432 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.913565s. I0423 01:37:56.844502 123736353146432 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 01:37:57.264251 123736353146432 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 01:37:57.264441 123736353146432 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.572881s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9/items I0423 01:37:57.265283 123736353146432 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9/items I0423 01:37:57.505921 123736353146432 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9 I0423 01:37:58.220222 123736353146432 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9`. I0423 01:37:58.220963 123736353146432 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 43.494511s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_06_grad_accum/checkpoints/9 I0423 01:37:58.221059 123736353146432 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.745631s. I0423 01:37:58.221271 123736183277120 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 01:37:58.221396 123736183277120 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 01:37:58.221448 123736183277120 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 01:37:58.221488 123736183277120 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 01:37:58.221658 123767550224192 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. I0423 01:37:58.221806 123767550224192 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 01:37:58.222794 123767550224192 metric_logger.py:196] completed step: 9, seconds: 0.422, TFLOP/s/device: 128.815, Tokens/s/device: 19416.417, 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