MaxView

XPK Start: Fri Apr 17 13:21:23 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 13:21:46.814213: 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)
I0417 13:21:46.994390 132498554505024 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 13:21:56.033740 132498554505024 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 13:21:56.036257 132498554505024 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-ei1mp-slice-job-0-0.mt-05-fp8-ei1mp:8482
I0417 13:21:57.426766 132498554505024 max_utils.py:284] Jax distributed system initialized!
I0417 13:22:03.444493 132498554505024 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 13:22:03.444594 132498554505024 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 13:22:03.444633 132498554505024 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0417 13:22:03.444669 132498554505024 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 13:22:04.469048 132498554505024 maxtext_utils.py:1548] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 13:22:04.469316 132498554505024 checkpointing.py:677] Setting up checkpoint logger...
I0417 13:22:04.469369 132498554505024 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 13:22:04.469410 132498554505024 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 13:22:04.469740 132498554505024 base_pytree_checkpoint_handler.py:441] 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 0x78810194f980>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 13:22:07.424211 132498554505024 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0417 13:22:07.424447 132498554505024 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x787ff5473890>}, handler_registry=None
I0417 13:22:07.424683 132498554505024 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x787ff5473890>` 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`.
I0417 13:22:07.424730 132498554505024 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787ff5470e60>` 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`.
I0417 13:22:07.424766 132498554505024 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 0x787ff5473890>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x787ff5473890>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787ff5470e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x787ff5470e60>}).
I0417 13:22:07.425108 132498554505024 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 13:22:07.425179 132498554505024 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x786d5472c040> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 13:22:08.527495 132498554505024 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints
I0417 13:22:09.402518 132498554505024 checkpoint_manager.py:929] [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_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, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x787ff545e120>
I0417 13:22:09.402685 132498554505024 checkpointing.py:301] Checkpoint manager created!
I0417 13:22:10.461959 132498554505024 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 13:22:10.462074 132498554505024 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:22:10.852912 132498554505024 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:22:10.852999 132498554505024 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:22:10.869051 132498554505024 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:22:10.869117 132498554505024 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:22:10.920375 132498554505024 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:22:10.920453 132498554505024 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:22:10.936542 132498554505024 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:22:10.936601 132498554505024 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:22:10.952814 132498554505024 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:22:10.952888 132498554505024 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:22:10.977310 132498554505024 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 13:22:10.977377 132498554505024 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:22:11.027914 132498554505024 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 13:22:11.027984 132498554505024 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 13:22:11.463059 132498554505024 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 13:22:11.463167 132498554505024 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0417 13:22:13.383851 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.383969 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384015 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384053 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384092 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384125 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384156 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384185 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384214 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384243 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384272 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384300 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384327 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384354 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384381 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384407 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384435 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384463 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384490 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384516 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384543 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384569 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384596 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384622 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384649 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384674 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384700 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384726 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384760 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384788 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384814 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384873 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384909 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384938 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384966 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.384993 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385020 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385046 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385072 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385104 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385131 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history
    Shape:     float32[16,1024]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385157 132498554505024 maxtext_utils.py:1651]  params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale
    Shape:     float32[16,1]
    Logical:   P()
    Physical:  ()
I0417 13:22:13.385205 132498554505024 maxtext_utils.py:1651]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 13:22:13.385260 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 13:22:13.385298 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 13:22:13.385344 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   P('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 13:22:13.385389 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 13:22:13.385420 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 13:22:13.385465 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0417 13:22:13.385511 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   P('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0417 13:22:13.385546 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0417 13:22:13.385578 132498554505024 maxtext_utils.py:1651]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0417 13:22:13.385621 132498554505024 maxtext_utils.py:1651]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0417 13:22:13.385662 132498554505024 maxtext_utils.py:1651]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0417 13:22:15.012288 132498554505024 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:22:15.012382 132498554505024 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:22:15.027616 132498554505024 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:22:15.027673 132498554505024 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:22:28.395267 132498554505024 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0417 13:22:28.402760 132498554505024 metric_logger.py:301] number parameters: 1.105 billion
I0417 13:22:43.710010 132498554505024 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0417 13:22:43.877583 132498554505024 checkpointing.py:776] Waited 0.16755938529968262 seconds for step 0 to finish before starting checkpointing.
I0417 13:22:43.880004 132498554505024 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:22:43.882045 132498554505024 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0
I0417 13:22:43.883417 132498554505024 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0.
I0417 13:22:45.075685 132498554505024 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 13:22:45.094344 132498554505024 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array.
I0417 13:22:45.094467 132498554505024 replica_slices.py:424] 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
I0417 13:22:45.326382 132372292953856 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0
I0417 13:22:45.497108 132498554505024 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.403543s
I0417 13:22:45.497986 132498554505024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 3.732 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.413360595703125 s) (per-host)
I0417 13:22:45.498082 132498554505024 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.413480s (batch_requests_ready=0.003467s, total_serialization_initiated=0.409202s, others=0.000811s)
I0417 13:22:45.498198 132498554505024 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.417954s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.417936)
I0417 13:22:45.498984 132498554505024 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.62 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0.
I0417 13:22:45.499275 132371194033920 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 13:42:45.499238
I0417 13:22:45.517221 132498554505024 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:22:45.517494 132372301346560 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:22:45.517620 132498554505024 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432163.8799853, 'wait_for_prev_duration_secs': 6.175041198730469e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432163.8820834, 'checkpointer_blocking_duration_secs': 1.6173536777496338, 'get_old_steps_start_time': 1776432165.4994628, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776432163.8781085, 'checkpoint_manager_blocking_duration_secs': 1.6394739151000977}
I0417 13:22:45.517727 132498554505024 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0417 13:22:45.517778 132498554505024 max_utils.py:750] 
Memstats: After params initialized:
I0417 13:22:45.517830 132498554505024 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_0(process=0,(0,0,0,0))
I0417 13:22:45.517884 132498554505024 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_1(process=0,(1,0,0,0))
I0417 13:22:45.517913 132498554505024 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_4(process=0,(0,1,0,0))
I0417 13:22:45.517938 132498554505024 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_5(process=0,(1,1,0,0))
I0417 13:22:45.874782 132498554505024 metric_logger.py:196] completed step: 0, seconds: 14.888, TFLOP/s/device: 0.913, Tokens/s/device: 137.561, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52776.805
I0417 13:22:45.878202 132498554505024 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/tensorboard/'
I0417 13:22:46.304766 132498554505024 metric_logger.py:196] completed step: 1, seconds: 2.162, TFLOP/s/device: 6.284, Tokens/s/device: 947.158, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52761.840
I0417 13:22:46.461271 132498554505024 metric_logger.py:196] completed step: 2, seconds: 0.431, TFLOP/s/device: 31.519, Tokens/s/device: 4750.847, total_weights: 65536, loss: 10.816, lm_loss: 10.816, perplexity: 49832.398
I0417 13:22:46.619384 132498554505024 metric_logger.py:196] completed step: 3, seconds: 0.010, TFLOP/s/device: 1379.964, Tokens/s/device: 208003.250, total_weights: 65536, loss: 10.431, lm_loss: 10.431, perplexity: 33901.820
I0417 13:22:47.526977 132371227604736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/items
I0417 13:22:47.703940 132370690733824 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432167280665812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776432167280665812, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:22:49.082156    2821 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 13:22:51.275780 132371210819328 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/items/array_metadatas/process_0
I0417 13:23:10.404405 132371202426624 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 21.330975s (commit=18.960584s, array_metadata_write=2.370391s)
I0417 13:23:10.405563 132371194033920 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 62.390 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 25.32091736793518 s) (per-host)
I0417 13:23:10.405688 132371194033920 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 24.906297s.
I0417 13:23:11.068689 132371194033920 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432167280665812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:23:11.301432 132498554505024 metric_logger.py:196] completed step: 4, seconds: 0.158, TFLOP/s/device: 86.203, Tokens/s/device: 12993.522, total_weights: 65536, loss: 9.992, lm_loss: 9.992, perplexity: 21847.639
I0417 13:23:11.310577 132498554505024 metric_logger.py:196] completed step: 5, seconds: 0.158, TFLOP/s/device: 85.992, Tokens/s/device: 12961.615, total_weights: 65536, loss: 9.549, lm_loss: 9.549, perplexity: 14035.006
I0417 13:23:11.467056 132498554505024 metric_logger.py:196] completed step: 6, seconds: 24.680, TFLOP/s/device: 0.551, Tokens/s/device: 82.981, total_weights: 65536, loss: 9.111, lm_loss: 9.111, perplexity: 9058.688
I0417 13:23:11.625540 132498554505024 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1907.501, Tokens/s/device: 287519.304, total_weights: 65536, loss: 8.685, lm_loss: 8.685, perplexity: 5914.720
I0417 13:23:11.783939 132498554505024 metric_logger.py:196] completed step: 8, seconds: 0.157, TFLOP/s/device: 86.322, Tokens/s/device: 13011.353, total_weights: 65536, loss: 8.281, lm_loss: 8.281, perplexity: 3950.010
I0417 13:23:11.791165 132498554505024 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0417 13:23:12.100615 132498554505024 checkpointing.py:776] Waited 0.30941224098205566 seconds for step 10 to finish before starting checkpointing.
I0417 13:23:12.103289 132498554505024 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:23:12.204412 132371194033920 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000457s.
I0417 13:23:12.271901 132370690733824 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432167280665812, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:23:12.595990 132371194033920 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 13:23:13.516870 132371194033920 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.323525s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/items
I0417 13:23:13.518566 132371194033920 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0/items
I0417 13:23:14.049204 132371194033920 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0
I0417 13:23:15.508969 132371194033920 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0`.
I0417 13:23:15.509820 132371194033920 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 31.63 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/0
I0417 13:23:15.511220 132371194033920 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 30.011827s.
I0417 13:23:15.511434 132372301346560 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:23:15.511553 132372301346560 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:23:15.511616 132372301346560 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:23:15.514449 132372301346560 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0417 13:23:15.514625 132498554505024 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0417 13:23:15.514762 132498554505024 checkpoint_manager.py:1452] Waiting for previous save to complete took 3.411475 seconds. If this number is high, consider checkpointing less frequently.
I0417 13:23:15.516541 132498554505024 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10
I0417 13:23:15.518486 132498554505024 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10.
I0417 13:23:16.240148 132498554505024 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array.
I0417 13:23:16.240247 132498554505024 replica_slices.py:424] 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
I0417 13:23:16.280041 132498554505024 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040649s
I0417 13:23:16.280822 132498554505024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 32.014 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04818916320800781 s) (per-host)
I0417 13:23:16.280912 132498554505024 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048299s (batch_requests_ready=0.003166s, total_serialization_initiated=0.044427s, others=0.000706s)
I0417 13:23:16.281007 132498554505024 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.052274s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.052257)
I0417 13:23:16.281677 132498554505024 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10.
I0417 13:23:16.281955 132372301346560 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 13:43:16.281927
I0417 13:23:16.283937 132498554505024 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:23:16.284228 132369600206592 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:23:16.284362 132498554505024 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432192.1032546, 'wait_for_prev_duration_secs': 3.411475419998169, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432195.51658, 'checkpointer_blocking_duration_secs': 0.765491247177124, 'get_old_steps_start_time': 1776432196.282096, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776432192.1013732, 'checkpoint_manager_blocking_duration_secs': 4.182955741882324}
I0417 13:23:16.284518 132498554505024 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0417 13:23:16.285169 132498554505024 metric_logger.py:196] completed step: 9, seconds: 0.159, TFLOP/s/device: 85.643, Tokens/s/device: 12909.082, total_weights: 65536, loss: 7.908, lm_loss: 7.908, perplexity: 2720.116
I0417 13:23:16.295732 132498554505024 metric_logger.py:196] completed step: 10, seconds: 0.159, TFLOP/s/device: 85.704, Tokens/s/device: 12918.283, total_weights: 65536, loss: 7.568, lm_loss: 7.568, perplexity: 1936.198
I0417 13:23:16.450873 132498554505024 metric_logger.py:196] completed step: 11, seconds: 4.502, TFLOP/s/device: 3.018, Tokens/s/device: 454.886, total_weights: 65536, loss: 7.288, lm_loss: 7.288, perplexity: 1463.272
I0417 13:23:16.475490 132371194033920 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10
I0417 13:23:16.609182 132498554505024 metric_logger.py:196] completed step: 12, seconds: 0.009, TFLOP/s/device: 1531.289, Tokens/s/device: 230812.577, total_weights: 65536, loss: 7.032, lm_loss: 7.032, perplexity: 1132.061
I0417 13:23:17.103406 132498554505024 metric_logger.py:196] completed step: 13, seconds: 0.156, TFLOP/s/device: 87.011, Tokens/s/device: 13115.174, total_weights: 65536, loss: 6.815, lm_loss: 6.815, perplexity: 911.031
I0417 13:23:17.260040 132498554505024 metric_logger.py:196] completed step: 14, seconds: 0.644, TFLOP/s/device: 21.099, Tokens/s/device: 3180.262, total_weights: 65536, loss: 6.635, lm_loss: 6.635, perplexity: 761.080
I0417 13:23:17.418324 132498554505024 metric_logger.py:196] completed step: 15, seconds: 0.007, TFLOP/s/device: 1868.674, Tokens/s/device: 281666.896, total_weights: 65536, loss: 6.492, lm_loss: 6.492, perplexity: 659.915
I0417 13:23:17.576502 132498554505024 metric_logger.py:196] completed step: 16, seconds: 0.158, TFLOP/s/device: 86.076, Tokens/s/device: 12974.343, total_weights: 65536, loss: 6.380, lm_loss: 6.380, perplexity: 589.948
I0417 13:23:17.734804 132498554505024 metric_logger.py:196] completed step: 17, seconds: 0.158, TFLOP/s/device: 86.093, Tokens/s/device: 12976.809, total_weights: 65536, loss: 6.293, lm_loss: 6.293, perplexity: 540.941
I0417 13:23:17.893049 132498554505024 metric_logger.py:196] completed step: 18, seconds: 0.158, TFLOP/s/device: 85.856, Tokens/s/device: 12941.139, total_weights: 65536, loss: 6.223, lm_loss: 6.223, perplexity: 504.108
I0417 13:23:18.050670 132498554505024 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0417 13:23:18.051710 132498554505024 checkpointing.py:776] Waited 0.0010600090026855469 seconds for step 19 to finish before starting checkpointing.
I0417 13:23:18.053820 132498554505024 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:23:18.598484 132371227604736 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10/items
I0417 13:23:22.351143 132370098857728 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10/items/array_metadatas/process_0
I0417 13:24:01.607232 132369616992000 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.451220s (commit=40.226691s, array_metadata_write=1.224529s)
I0417 13:24:01.608407 132372301346560 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.815 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.37574100494385 s) (per-host)
I0417 13:24:01.608529 132372301346560 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.326482s.
I0417 13:24:03.138863 132372301346560 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000435s.
I0417 13:24:03.506199 132372301346560 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 13:24:04.558704 132372301346560 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.056906s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10/items
I0417 13:24:04.561614 132372301346560 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10/items
I0417 13:24:05.071828 132372301346560 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10
I0417 13:24:06.479183 132372301346560 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10`.
I0417 13:24:06.480024 132372301346560 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.96 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/10
I0417 13:24:06.481708 132372301346560 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.199658s.
I0417 13:24:06.481954 132369600206592 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:24:06.482073 132369600206592 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:24:06.482135 132369600206592 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:24:06.484117 132369600206592 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0417 13:24:06.484278 132498554505024 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0417 13:24:06.484398 132498554505024 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.430585 seconds. If this number is high, consider checkpointing less frequently.
I0417 13:24:06.486174 132498554505024 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19
I0417 13:24:06.488075 132498554505024 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19.
I0417 13:24:07.203676 132498554505024 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array.
I0417 13:24:07.203772 132498554505024 replica_slices.py:424] 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
I0417 13:24:07.242822 132498554505024 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.039944s
I0417 13:24:07.243636 132498554505024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 32.584 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04734611511230469 s) (per-host)
I0417 13:24:07.243705 132498554505024 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.047435s (batch_requests_ready=0.003049s, total_serialization_initiated=0.043637s, others=0.000750s)
I0417 13:24:07.243780 132498554505024 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051418s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=0.051405)
I0417 13:24:07.244506 132498554505024 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19.
I0417 13:24:07.244890 132371177260800 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 13:44:07.244847
I0417 13:24:07.247066 132498554505024 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:24:07.247413 132369600206592 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:24:07.247550 132498554505024 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432198.0537906, 'wait_for_prev_duration_secs': 48.43058514595032, 'time_between_consecutive_saves_sec': 2.539302349090576, 'checkpointer_blocking_start_time': 1776432246.4862142, 'checkpointer_blocking_duration_secs': 0.7588226795196533, 'get_old_steps_start_time': 1776432247.2450554, 'get_old_steps_duration_secs': 2.3126602172851562e-05, 'checkpoint_manager_blocking_start_time': 1776432198.0520124, 'checkpoint_manager_blocking_duration_secs': 49.195504903793335}
I0417 13:24:07.247719 132498554505024 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0417 13:24:07.247763 132498554505024 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:24:07.441231 132372301346560 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19
I0417 13:24:09.154053 132371210819328 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19/items
I0417 13:24:13.354792 132371227604736 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19/items/array_metadatas/process_0
I0417 13:24:52.016001 132371194046208 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.711854s (commit=40.076078s, array_metadata_write=1.635776s)
I0417 13:24:52.017180 132371177260800 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.246 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.820860385894775 s) (per-host)
I0417 13:24:52.017308 132371177260800 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.772298s.
I0417 13:24:53.865651 132371177260800 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000433s.
I0417 13:24:54.236896 132371177260800 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 13:24:55.171315 132371177260800 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.293360s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19/items
I0417 13:24:55.173023 132371177260800 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19/items
I0417 13:24:55.692875 132371177260800 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19
I0417 13:24:57.060439 132371177260800 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19`.
I0417 13:24:57.061273 132371177260800 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.58 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_05_fp8/checkpoints/19
I0417 13:24:57.062779 132371177260800 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.817766s.
I0417 13:24:57.063019 132369600206592 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:24:57.063144 132369600206592 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:24:57.063196 132369600206592 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:24:57.065231 132369600206592 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0417 13:24:57.065397 132498554505024 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0417 13:24:57.065546 132498554505024 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:24:57.066730 132498554505024 metric_logger.py:196] completed step: 19, seconds: 0.158, TFLOP/s/device: 85.843, Tokens/s/device: 12939.177, total_weights: 65536, loss: 6.167, lm_loss: 6.167, perplexity: 476.947
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Fri Apr 17 13:25:09 UTC 2026
EXIT_CODE=0