2026-04-22 17:56:40.205673: 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 00:00:1776880600.218998 2029305 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776880600.222840 2029305 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776880600.234377 2029305 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880600.234393 2029305 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880600.234395 2029305 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776880600.234396 2029305 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-22 17:56:43.915471: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0422 17:56:44.052324 136045651191616 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0422 17:57:14.639053 136045651191616 max_utils.py:800] System Information: Jax Version: 0.8.1 I0422 17:57:14.639326 136045651191616 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0422 17:57:14.639375 136045651191616 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0422 17:57:14.639405 136045651191616 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0422 17:57:14.639588 136045651191616 train.py:706] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0422 17:57:15.299570 136045651191616 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 17:57:15.299853 136045651191616 checkpointing.py:677] Setting up checkpoint logger... I0422 17:57:15.299898 136045651191616 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 17:57:15.299936 136045651191616 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0422 17:57:15.300381 136045651191616 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 0x7bbb51baef90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 17:57:17.731302 136045651191616 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0422 17:57:17.732516 136045651191616 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 0x7bb9f31c1310>}, handler_registry=None I0422 17:57:17.732883 136045651191616 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bb9f31c1310>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0422 17:57:17.732930 136045651191616 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb97635e180>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0422 17:57:17.732962 136045651191616 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 0x7bb9f31c1310>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bb9f31c1310>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb97635e180>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb97635e180>}). I0422 17:57:17.733283 136045651191616 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0422 17:57:17.733337 136045651191616 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7bb976370540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0422 17:57:17.848449 136045651191616 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints I0422 17:57:17.848770 136045651191616 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bb97635e030> I0422 17:57:17.849084 136045651191616 checkpointing.py:301] Checkpoint manager created! I0422 17:57:17.944276 136045651191616 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0422 17:57:17.947147 136045651191616 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0422 17:57:18.006196 136045651191616 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0422 17:57:18.028653 136045651191616 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0422 17:57:18.028723 136045651191616 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0422 17:57:18.795493 136045651191616 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 17:57:18.795598 136045651191616 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 17:57:19.100453 136045651191616 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:57:19.100546 136045651191616 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 17:57:19.113250 136045651191616 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 17:57:19.113302 136045651191616 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0422 17:57:19.163143 136045651191616 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:57:19.163222 136045651191616 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 17:57:19.176178 136045651191616 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:57:19.176232 136045651191616 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 17:57:19.188922 136045651191616 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 17:57:19.188981 136045651191616 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 17:57:19.208938 136045651191616 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 17:57:19.209005 136045651191616 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0422 17:57:19.258621 136045651191616 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 17:57:19.258696 136045651191616 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0422 17:57:19.914373 136045651191616 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0422 17:57:19.914480 136045651191616 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0422 17:57:20.983957 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984092 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984134 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984165 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984192 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984217 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984241 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984264 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984288 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984311 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984333 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984354 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984376 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984397 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984418 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984439 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984458 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984479 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984498 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984519 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984539 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984559 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984579 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984599 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984619 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984639 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984660 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984679 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984699 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984720 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984742 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984763 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984783 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984802 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984822 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984843 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984863 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984883 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984903 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984922 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.984941 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0422 17:57:20.984961 136045651191616 maxtext_utils.py:1707] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0422 17:57:20.985011 136045651191616 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0422 17:57:20.985064 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 17:57:20.985094 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 17:57:20.985130 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0422 17:57:20.985163 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 17:57:20.985187 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0422 17:57:20.985222 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 17:57:20.985258 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0422 17:57:20.985284 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0422 17:57:20.985308 136045651191616 maxtext_utils.py:1707] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0422 17:57:20.985341 136045651191616 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 17:57:20.985373 136045651191616 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 17:57:22.438199 136045651191616 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:57:22.438293 136045651191616 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0422 17:57:22.450687 136045651191616 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0422 17:57:22.450739 136045651191616 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0422 17:57:23.854772 136045651191616 max_utils.py:791] Total memory size: 3.5 GB, Output size: 1.5 GB, Temp size: 1.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0422 17:57:23.855784 136045651191616 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0422 17:57:23.856159 136045651191616 metric_logger.py:301] number parameters: 1.105 billion I0422 17:57:30.094520 136045651191616 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0422 17:57:30.212526 136045651191616 checkpointing.py:776] Waited 0.11798715591430664 seconds for step 0 to finish before starting checkpointing. I0422 17:57:30.217481 136045651191616 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:57:30.217674 136045651191616 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0422 17:57:30.218107 136045651191616 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0. I0422 17:57:30.307667 136045651191616 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0422 17:57:30.391829 135920034514496 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0 I0422 17:57:30.401696 136045651191616 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0422 17:57:30.401802 136045651191616 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0422 17:57:31.093454 135920024028736 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/items I0422 17:57:31.238777 135919854159424 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880650998123846, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776880650998123846, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0422 17:57:31.615056 2031945 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0422 17:57:31.615116 2031945 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0422 17:57:33.873619 136045651191616 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.472346s I0422 17:57:33.881481 136045651191616 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.455 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 17:57:33.881618 136045651191616 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.572858s (batch_requests_ready=0.086143s, total_serialization_initiated=3.478894s, others=0.007822s) I0422 17:57:33.881703 136045651191616 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.573458s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=3.573438) I0422 17:57:33.883322 135920045000256 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 17:57:33.883440 136045651191616 async_checkpointer.py:561] Finished blocking save. Time taken: 3.665723s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0. I0422 17:57:33.883576 136045651191616 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:57:33.883744 135919503935040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:57:33.883841 136045651191616 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880650.217466, 'wait_for_prev_duration_secs': 5.459785461425781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880650.2176986, 'checkpointer_blocking_duration_secs': 3.6658003330230713, 'get_old_steps_start_time': 1776880653.8835173, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776880650.2173705, 'checkpoint_manager_blocking_duration_secs': 3.6664445400238037} I0422 17:57:33.883954 136045651191616 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0422 17:57:33.883996 136045651191616 max_utils.py:750] Memstats: After params initialized: I0422 17:57:33.884062 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_0(process=0,(0,0,0,0)) I0422 17:57:33.884091 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_1(process=0,(1,0,0,0)) I0422 17:57:33.884111 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_2(process=0,(0,1,0,0)) I0422 17:57:33.884129 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_3(process=0,(1,1,0,0)) I0422 17:57:33.884146 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_4(process=0,(0,2,0,0)) I0422 17:57:33.884165 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_5(process=0,(1,2,0,0)) I0422 17:57:33.884182 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_6(process=0,(0,3,0,0)) I0422 17:57:33.884201 136045651191616 max_utils.py:756] Using (GB) 1.61 / 31.25 (5.152000%) on TPU_7(process=0,(1,3,0,0)) I0422 17:57:34.265957 136045651191616 metric_logger.py:196] completed step: 0, seconds: 6.238, TFLOP/s/device: 2.178, Tokens/s/device: 328.330, total_weights: 13328, loss: 10.844, lm_loss: 10.844, perplexity: 51212.484 I0422 17:57:34.267005 136045651191616 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/tensorboard/' I0422 17:57:34.375256 136045651191616 metric_logger.py:196] completed step: 1, seconds: 4.168, TFLOP/s/device: 3.260, Tokens/s/device: 491.334, total_weights: 12332, loss: 10.871, lm_loss: 10.871, perplexity: 52634.344 I0422 17:57:34.486984 136045651191616 metric_logger.py:196] completed step: 2, seconds: 0.012, TFLOP/s/device: 1096.621, Tokens/s/device: 165294.592, total_weights: 15161, loss: 10.871, lm_loss: 10.871, perplexity: 52617.816 I0422 17:57:34.598971 136045651191616 metric_logger.py:196] completed step: 3, seconds: 0.110, TFLOP/s/device: 123.359, Tokens/s/device: 18594.010, total_weights: 13327, loss: 10.857, lm_loss: 10.857, perplexity: 51888.457 I0422 17:57:34.638154 135919864645184 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/items/array_metadatas/process_0 I0422 17:57:54.562953 136045651191616 metric_logger.py:196] completed step: 4, seconds: 0.112, TFLOP/s/device: 121.407, Tokens/s/device: 18299.766, total_weights: 11939, loss: 10.861, lm_loss: 10.861, perplexity: 52094.562 I0422 17:57:54.660111 136045651191616 metric_logger.py:196] completed step: 5, seconds: 0.112, TFLOP/s/device: 121.224, Tokens/s/device: 18272.173, total_weights: 15502, loss: 10.882, lm_loss: 10.882, perplexity: 53230.324 I0422 17:57:54.798069 136045651191616 metric_logger.py:196] completed step: 6, seconds: 20.048, TFLOP/s/device: 0.678, Tokens/s/device: 102.157, total_weights: 13864, loss: 10.886, lm_loss: 10.886, perplexity: 53423.605 I0422 17:57:54.899631 136045651191616 metric_logger.py:196] completed step: 7, seconds: 0.088, TFLOP/s/device: 154.552, Tokens/s/device: 23295.758, total_weights: 12988, loss: 10.829, lm_loss: 10.829, perplexity: 50459.859 I0422 17:57:55.028482 136045651191616 metric_logger.py:196] completed step: 8, seconds: 0.103, TFLOP/s/device: 132.387, Tokens/s/device: 19954.790, total_weights: 13820, loss: 10.848, lm_loss: 10.848, perplexity: 51456.266 I0422 17:57:55.074336 136045651191616 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0422 17:57:55.232517 136045651191616 checkpointing.py:776] Waited 0.15813970565795898 seconds for step 10 to finish before starting checkpointing. I0422 17:57:55.234344 136045651191616 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:58:20.436107 135919705261632 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 46.553778s (commit=46.059088s, array_metadata_write=0.494690s) I0422 17:58:20.437316 135920045000256 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 252.114 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 50 seconds) (per-host) I0422 17:58:20.437388 135920045000256 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.553976s. I0422 17:58:20.688876 135920045000256 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880650998123846, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0422 17:58:20.903407 135920045000256 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 17:58:21.099029 135919854159424 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776880650998123846, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0422 17:58:21.285259 135920045000256 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 17:58:21.285555 135920045000256 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.556081s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/items I0422 17:58:21.286297 135920045000256 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0/items I0422 17:58:21.531781 135920045000256 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0 I0422 17:58:22.226302 135920045000256 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0`. I0422 17:58:22.227072 135920045000256 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 52.009358s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/0 I0422 17:58:22.227164 135920045000256 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 48.343758s. I0422 17:58:22.227420 135919503935040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:58:22.227549 135919503935040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:58:22.227616 135919503935040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:58:22.227671 135919503935040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 17:58:22.227870 136045651191616 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0422 17:58:22.228021 136045651191616 checkpoint_manager.py:1441] Waiting for previous save to complete took 26.993713 seconds. If this number is high, consider checkpointing less frequently. I0422 17:58:22.229470 136045651191616 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 10 I0422 17:58:22.229828 136045651191616 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10. I0422 17:58:22.410766 135920045000256 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10 I0422 17:58:22.418168 136045651191616 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0422 17:58:22.418277 136045651191616 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0422 17:58:23.185243 135919864645184 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10/items I0422 17:58:24.941012 136045651191616 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 2.523367s I0422 17:58:24.948243 136045651191616 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.706 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 2 seconds) (per-host) I0422 17:58:24.948316 136045651191616 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 2.622747s (batch_requests_ready=0.085818s, total_serialization_initiated=2.529860s, others=0.007070s) I0422 17:58:24.948802 136045651191616 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 2.623781s (all_items=0.000025s, per_item={'items': '0.00002456'}, temp_paths=2.623757) I0422 17:58:24.950492 135920045000256 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 17:58:24.950622 136045651191616 async_checkpointer.py:561] Finished blocking save. Time taken: 2.721098s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10. I0422 17:58:24.950791 136045651191616 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:58:24.950986 135919503935040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:58:24.951113 136045651191616 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880675.2342765, 'wait_for_prev_duration_secs': 26.993712902069092, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776880702.2294993, 'checkpointer_blocking_duration_secs': 2.721182346343994, 'get_old_steps_start_time': 1776880704.9507031, 'get_old_steps_duration_secs': 3.910064697265625e-05, 'checkpoint_manager_blocking_start_time': 1776880675.2341807, 'checkpoint_manager_blocking_duration_secs': 29.716907024383545} I0422 17:58:24.951245 136045651191616 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0422 17:58:24.951950 136045651191616 metric_logger.py:196] completed step: 9, seconds: 0.121, TFLOP/s/device: 111.996, Tokens/s/device: 16881.254, total_weights: 12300, loss: 10.857, lm_loss: 10.857, perplexity: 51900.086 I0422 17:58:24.963283 136045651191616 metric_logger.py:196] completed step: 10, seconds: 0.105, TFLOP/s/device: 129.543, Tokens/s/device: 19526.148, total_weights: 13436, loss: 10.876, lm_loss: 10.876, perplexity: 52905.383 I0422 17:58:25.071993 136045651191616 metric_logger.py:196] completed step: 11, seconds: 29.886, TFLOP/s/device: 0.455, Tokens/s/device: 68.527, total_weights: 13817, loss: 10.853, lm_loss: 10.853, perplexity: 51714.605 I0422 17:58:25.183681 136045651191616 metric_logger.py:196] completed step: 12, seconds: 0.010, TFLOP/s/device: 1299.706, Tokens/s/device: 195905.873, total_weights: 12076, loss: 10.852, lm_loss: 10.852, perplexity: 51622.902 I0422 17:58:25.295787 136045651191616 metric_logger.py:196] completed step: 13, seconds: 0.109, TFLOP/s/device: 124.105, Tokens/s/device: 18706.442, total_weights: 11048, loss: 10.864, lm_loss: 10.864, perplexity: 52261.953 I0422 17:58:25.407765 136045651191616 metric_logger.py:196] completed step: 14, seconds: 0.112, TFLOP/s/device: 121.564, Tokens/s/device: 18323.507, total_weights: 14267, loss: 10.834, lm_loss: 10.834, perplexity: 50738.680 I0422 17:58:25.519543 136045651191616 metric_logger.py:196] completed step: 15, seconds: 0.112, TFLOP/s/device: 121.276, Tokens/s/device: 18280.002, total_weights: 13004, loss: 10.832, lm_loss: 10.832, perplexity: 50634.074 I0422 17:58:25.631438 136045651191616 metric_logger.py:196] completed step: 16, seconds: 0.112, TFLOP/s/device: 121.226, Tokens/s/device: 18272.499, total_weights: 13811, loss: 10.867, lm_loss: 10.867, perplexity: 52422.980 I0422 17:58:25.695780 136029078029888 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10/items/array_metadatas/process_0 I0422 17:58:25.743508 136045651191616 metric_logger.py:196] completed step: 17, seconds: 0.112, TFLOP/s/device: 121.580, Tokens/s/device: 18325.802, total_weights: 13709, loss: 10.850, lm_loss: 10.850, perplexity: 51525.906 I0422 17:58:25.855509 136045651191616 metric_logger.py:196] completed step: 18, seconds: 0.112, TFLOP/s/device: 121.487, Tokens/s/device: 18311.874, total_weights: 12831, loss: 10.839, lm_loss: 10.839, perplexity: 50989.945 I0422 17:58:25.966688 136045651191616 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0422 17:58:25.968474 136045651191616 checkpointing.py:776] Waited 0.0018088817596435547 seconds for step 19 to finish before starting checkpointing. I0422 17:58:25.968966 136045651191616 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:58:55.434819 135919864645184 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.485350s (commit=30.026665s, array_metadata_write=0.458685s) I0422 17:58:55.435981 135920045000256 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 381.696 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0422 17:58:55.436037 135920045000256 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.485456s. I0422 17:58:55.976781 135920045000256 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 17:58:56.400265 135920045000256 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 17:58:56.400568 135920045000256 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.624484s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10/items I0422 17:58:56.401445 135920045000256 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10/items I0422 17:58:56.632813 135920045000256 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10 I0422 17:58:57.319027 135920045000256 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10`. I0422 17:58:57.319900 135920045000256 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.090385s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/10 I0422 17:58:57.319996 135920045000256 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.369406s. I0422 17:58:57.320266 135919503935040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:58:57.320392 135919503935040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:58:57.320461 135919503935040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:58:57.320512 135919503935040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 17:58:57.320719 136045651191616 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0422 17:58:57.320858 136045651191616 checkpoint_manager.py:1441] Waiting for previous save to complete took 31.351899 seconds. If this number is high, consider checkpointing less frequently. I0422 17:58:57.321591 136045651191616 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 19 I0422 17:58:57.321961 136045651191616 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19. I0422 17:58:57.508314 135920045000256 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19 I0422 17:58:57.517002 136045651191616 jax_array_handlers.py:347] Scheduling D2H of 81 prioritized jax.Array. I0422 17:58:57.517130 136045651191616 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0422 17:58:58.241247 136029078029888 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19/items I0422 17:59:00.530504 136045651191616 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 3.013960s I0422 17:59:00.537855 136045651191616 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.948 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 3 seconds) (per-host) I0422 17:59:00.537919 136045651191616 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 3.126099s (batch_requests_ready=0.098073s, total_serialization_initiated=3.020791s, others=0.007235s) I0422 17:59:00.538006 136045651191616 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 3.126658s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=3.126641) I0422 17:59:00.540068 135919864645184 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0422 17:59:00.540155 136045651191616 async_checkpointer.py:561] Finished blocking save. Time taken: 3.218517s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19. I0422 17:59:00.540341 136045651191616 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 17:59:00.540498 135919503935040 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0422 17:59:00.540600 136045651191616 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776880705.9689374, 'wait_for_prev_duration_secs': 31.35189914703369, 'time_between_consecutive_saves_sec': 3.7412500381469727, 'checkpointer_blocking_start_time': 1776880737.3216202, 'checkpointer_blocking_duration_secs': 3.2186460494995117, 'get_old_steps_start_time': 1776880740.5402856, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1776880705.9688923, 'checkpoint_manager_blocking_duration_secs': 34.571683406829834} I0422 17:59:00.540741 136045651191616 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0422 17:59:00.540777 136045651191616 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 17:59:01.292921 135919833187904 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19/items/array_metadatas/process_0 I0422 17:59:30.941353 135919822702144 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.402750s (commit=29.917341s, array_metadata_write=0.485409s) I0422 17:59:30.942603 135919864645184 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 376.911 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 33 seconds) (per-host) I0422 17:59:30.942728 135919864645184 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 30.402527s. I0422 17:59:31.424164 135919864645184 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0422 17:59:31.844732 135919864645184 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0422 17:59:31.844951 135919864645184 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.585032s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19/items I0422 17:59:31.845737 135919864645184 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19/items I0422 17:59:32.097346 135919864645184 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19 I0422 17:59:32.788953 135919864645184 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19`. I0422 17:59:32.789710 135919864645184 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 35.468078s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260422_174509/linen_main_20260422_174509_05_fp8/checkpoints/19 I0422 17:59:32.789775 135919864645184 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 32.249576s. I0422 17:59:32.790007 135919503935040 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 17:59:32.790143 135919503935040 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0422 17:59:32.790201 135919503935040 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 17:59:32.790244 135919503935040 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 17:59:32.790387 136045651191616 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0422 17:59:32.790541 136045651191616 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 17:59:32.791485 136045651191616 metric_logger.py:196] completed step: 19, seconds: 0.112, TFLOP/s/device: 121.445, Tokens/s/device: 18305.491, total_weights: 13717, loss: 10.851, lm_loss: 10.851, perplexity: 51591.992 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops