XPK Start: Wed Apr 22 09:42:42 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. `rope_parameters`'s factor field must be a float >= 1, got 40 `rope_parameters`'s beta_fast field must be a float, got 32 `rope_parameters`'s beta_slow field must be a float, got 1 DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 2026-04-22 09:43:07.075792: 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 09:43:07.285386 133766386374464 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 09:43:16.325722 133766386374464 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 09:43:16.327962 133766386374464 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-356wo-slice-job-0-0.mt-05-fp8-356wo:8482 I0422 09:43:18.166003 133766386374464 max_utils.py:284] Jax distributed system initialized! I0422 09:43:24.182468 133766386374464 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 09:43:24.182573 133766386374464 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 09:43:24.182613 133766386374464 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 I0422 09:43:24.182662 133766386374464 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 09:43:24.871912 133766386374464 maxtext_utils.py:1732] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 09:43:24.872505 133766386374464 maxtext_utils.py:1732] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 09:43:24.872705 133766386374464 checkpointing.py:688] Setting up checkpoint logger... I0422 09:43:24.872758 133766386374464 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 09:43:24.872802 133766386374464 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 09:43:24.873147 133766386374464 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 0x79a86030e5d0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 09:43:27.709119 133766386374464 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 09:43:27.709359 133766386374464 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7994245266f0>}, handler_registry=None I0422 09:43:27.709600 133766386374464 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7994245266f0>` 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 09:43:27.709647 133766386374464 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7994245289b0>` 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 09:43:27.709693 133766386374464 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 0x7994245266f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7994245266f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7994245289b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7994245289b0>}). I0422 09:43:27.710017 133766386374464 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 09:43:27.710089 133766386374464 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7994042d9d00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 09:43:28.885050 133766386374464 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints I0422 09:43:29.301126 133766386374464 checkpoint_manager.py:929] [process=5][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_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x799424527680> I0422 09:43:29.301313 133766386374464 checkpointing.py:302] Checkpoint manager created! I0422 09:43:30.375786 133766386374464 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 09:43:30.375894 133766386374464 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 09:43:30.762250 133766386374464 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 09:43:30.762345 133766386374464 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 09:43:30.783530 133766386374464 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 09:43:30.783624 133766386374464 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 09:43:30.836339 133766386374464 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 09:43:30.836416 133766386374464 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 09:43:30.852682 133766386374464 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 09:43:30.852749 133766386374464 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 09:43:30.868961 133766386374464 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 09:43:30.869023 133766386374464 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 09:43:30.894349 133766386374464 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 09:43:30.894418 133766386374464 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 09:43:30.950061 133766386374464 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 09:43:30.950164 133766386374464 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 09:43:31.391822 133766386374464 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 09:43:31.391934 133766386374464 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0422 09:43:33.307971 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308091 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308139 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308176 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308211 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308245 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308277 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308308 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308337 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308366 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308395 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308427 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308458 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308486 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308514 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308541 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308569 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308600 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308631 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308677 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308707 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308735 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308762 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308789 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308815 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308842 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.308869 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.308967 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309035 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309077 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309113 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309145 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309175 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309213 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309242 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309269 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309297 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309324 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309352 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309380 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309407 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0422 09:43:33.309434 133766386374464 maxtext_utils.py:1835] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0422 09:43:33.309487 133766386374464 maxtext_utils.py:1835] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0422 09:43:33.309541 133766386374464 maxtext_utils.py:1835] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 09:43:33.309578 133766386374464 maxtext_utils.py:1835] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0422 09:43:33.309626 133766386374464 maxtext_utils.py:1835] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0422 09:43:33.309700 133766386374464 maxtext_utils.py:1835] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0422 09:43:33.309746 133766386374464 maxtext_utils.py:1835] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0422 09:43:33.309796 133766386374464 maxtext_utils.py:1835] 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) I0422 09:43:33.309844 133766386374464 maxtext_utils.py:1835] 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') I0422 09:43:33.309880 133766386374464 maxtext_utils.py:1835] 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) I0422 09:43:33.309913 133766386374464 maxtext_utils.py:1835] 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) I0422 09:43:33.309959 133766386374464 maxtext_utils.py:1835] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0422 09:43:33.310003 133766386374464 maxtext_utils.py:1835] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0422 09:43:34.932264 133766386374464 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 09:43:34.932356 133766386374464 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0422 09:43:34.947734 133766386374464 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0422 09:43:34.947792 133766386374464 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0422 09:43:48.859718 133766386374464 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. I0422 09:43:48.860668 133766386374464 metric_logger.py:301] number parameters: 1.105 billion I0422 09:44:04.693217 133766386374464 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0422 09:44:04.865704 133766386374464 checkpointing.py:798] Waited 0.17245054244995117 seconds for step 0 to finish before starting checkpointing. I0422 09:44:04.868070 133766386374464 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 09:44:04.870038 133766386374464 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 0 I0422 09:44:04.871512 133766386374464 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0. I0422 09:44:05.593516 133766386374464 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 09:44:05.594414 133766386374464 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0422 09:44:05.594527 133766386374464 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 I0422 09:44:06.004378 133766386374464 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.410978s I0422 09:44:06.004548 133766386374464 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 3.679 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.4193687438964844 s) (per-host) I0422 09:44:06.004601 133766386374464 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.419432s (batch_requests_ready=0.003607s, total_serialization_initiated=0.415754s, others=0.000071s) I0422 09:44:06.004696 133766386374464 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.423552s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.423533) I0422 09:44:06.005504 133766386374464 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.14 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0. I0422 09:44:06.005806 133636989163264 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 10:04:06.005773 I0422 09:44:06.012799 133766386374464 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 09:44:06.013071 133635952862976 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0422 09:44:06.013204 133766386374464 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776851044.868052, 'wait_for_prev_duration_secs': 5.984306335449219e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776851044.8700743, 'checkpointer_blocking_duration_secs': 1.1358966827392578, 'get_old_steps_start_time': 1776851046.0059974, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776851044.8662066, 'checkpoint_manager_blocking_duration_secs': 1.1469531059265137} I0422 09:44:06.013317 133766386374464 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0422 09:44:06.013369 133766386374464 max_utils.py:750] Memstats: After params initialized: I0422 09:44:06.013423 133766386374464 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_18(process=5,(2,4,0,0)) I0422 09:44:06.013464 133766386374464 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_19(process=5,(3,4,0,0)) I0422 09:44:06.013498 133766386374464 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_22(process=5,(2,5,0,0)) I0422 09:44:06.013525 133766386374464 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_23(process=5,(3,5,0,0)) I0422 09:44:06.324776 133766386374464 metric_logger.py:196] completed step: 0, seconds: 15.832, TFLOP/s/device: 0.858, Tokens/s/device: 129.355, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52776.805 I0422 09:44:06.513396 133766386374464 metric_logger.py:196] completed step: 1, seconds: 1.629, TFLOP/s/device: 8.338, Tokens/s/device: 1256.828, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52761.840 I0422 09:44:06.921889 133766386374464 metric_logger.py:196] completed step: 2, seconds: 0.031, TFLOP/s/device: 435.862, Tokens/s/device: 65697.880, total_weights: 65536, loss: 10.816, lm_loss: 10.816, perplexity: 49832.398 I0422 09:44:07.080253 133766386374464 metric_logger.py:196] completed step: 3, seconds: 0.408, TFLOP/s/device: 33.264, Tokens/s/device: 5013.979, total_weights: 65536, loss: 10.431, lm_loss: 10.431, perplexity: 33901.820 I0422 09:44:07.397805 133766386374464 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.702, Tokens/s/device: 12465.762, total_weights: 65536, loss: 9.992, lm_loss: 9.992, perplexity: 21847.639 I0422 09:44:07.404385 133766386374464 metric_logger.py:196] completed step: 5, seconds: 0.158, TFLOP/s/device: 85.832, Tokens/s/device: 12937.461, total_weights: 65536, loss: 9.549, lm_loss: 9.549, perplexity: 14035.006 I0422 09:44:09.228236 2886 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 09:44:11.177553 133635961255680 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0/items/array_metadatas/process_5 I0422 09:44:25.768139 133636989163264 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 78.272 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 20.18291473388672 s) (per-host) I0422 09:44:25.768264 133636989163264 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 19.762342s. I0422 09:44:31.825897 133766386374464 metric_logger.py:196] completed step: 6, seconds: 0.318, TFLOP/s/device: 42.715, Tokens/s/device: 6438.510, total_weights: 65536, loss: 9.111, lm_loss: 9.111, perplexity: 9058.688 I0422 09:44:31.984368 133766386374464 metric_logger.py:196] completed step: 7, seconds: 24.264, TFLOP/s/device: 0.560, Tokens/s/device: 84.404, total_weights: 65536, loss: 8.685, lm_loss: 8.685, perplexity: 5914.720 I0422 09:44:32.142887 133766386374464 metric_logger.py:196] completed step: 8, seconds: 0.164, TFLOP/s/device: 83.007, Tokens/s/device: 12511.684, total_weights: 65536, loss: 8.281, lm_loss: 8.281, perplexity: 3950.010 I0422 09:44:32.147870 133766386374464 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0422 09:44:32.459325 133766386374464 checkpointing.py:798] Waited 0.311420202255249 seconds for step 10 to finish before starting checkpointing. I0422 09:44:32.461995 133766386374464 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 09:44:34.042226 133636989163264 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 28.036287s. I0422 09:44:34.042521 133635952862976 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 09:44:34.042645 133635952862976 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0422 09:44:34.042718 133635952862976 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 09:44:34.044483 133635952862976 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 09:44:34.044645 133766386374464 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0422 09:44:34.044786 133766386374464 checkpoint_manager.py:1452] Waiting for previous save to complete took 1.582793 seconds. If this number is high, consider checkpointing less frequently. I0422 09:44:34.046568 133766386374464 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 10 I0422 09:44:34.048596 133766386374464 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10. I0422 09:44:34.341128 133766386374464 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0422 09:44:34.341224 133766386374464 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 I0422 09:44:34.394441 133766386374464 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054223s I0422 09:44:34.394607 133766386374464 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 25.458 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.060598134994506836 s) (per-host) I0422 09:44:34.394670 133766386374464 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.060657s (batch_requests_ready=0.003279s, total_serialization_initiated=0.057311s, others=0.000067s) I0422 09:44:34.394777 133766386374464 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.064810s (all_items=0.000015s, per_item={'items': '0.00001478'}, temp_paths=0.064795) I0422 09:44:34.395482 133766386374464 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10. I0422 09:44:34.395775 133635952862976 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 10:04:34.395741 I0422 09:44:34.398108 133766386374464 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0422 09:44:34.398344 133633805379328 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0422 09:44:34.398483 133766386374464 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776851072.4619646, 'wait_for_prev_duration_secs': 1.5827932357788086, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776851074.0466065, 'checkpointer_blocking_duration_secs': 0.349315881729126, 'get_old_steps_start_time': 1776851074.3959465, 'get_old_steps_duration_secs': 3.3855438232421875e-05, 'checkpoint_manager_blocking_start_time': 1776851072.460153, 'checkpoint_manager_blocking_duration_secs': 1.938284158706665} I0422 09:44:34.398627 133766386374464 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0422 09:44:34.399398 133766386374464 metric_logger.py:196] completed step: 9, seconds: 0.158, TFLOP/s/device: 85.925, Tokens/s/device: 12951.615, total_weights: 65536, loss: 7.908, lm_loss: 7.908, perplexity: 2720.116 I0422 09:44:34.406817 133766386374464 metric_logger.py:196] completed step: 10, seconds: 0.158, TFLOP/s/device: 85.756, Tokens/s/device: 12926.029, total_weights: 65536, loss: 7.568, lm_loss: 7.568, perplexity: 1936.198 I0422 09:44:34.564582 133766386374464 metric_logger.py:196] completed step: 11, seconds: 2.257, TFLOP/s/device: 6.021, Tokens/s/device: 907.544, total_weights: 65536, loss: 7.288, lm_loss: 7.288, perplexity: 1463.272 I0422 09:44:35.121850 133766386374464 metric_logger.py:196] completed step: 12, seconds: 0.007, TFLOP/s/device: 2053.057, Tokens/s/device: 309459.051, total_weights: 65536, loss: 7.032, lm_loss: 7.032, perplexity: 1132.061 I0422 09:44:35.280403 133766386374464 metric_logger.py:196] completed step: 13, seconds: 0.553, TFLOP/s/device: 24.589, Tokens/s/device: 3706.345, total_weights: 65536, loss: 6.815, lm_loss: 6.815, perplexity: 911.031 I0422 09:44:35.438688 133766386374464 metric_logger.py:196] completed step: 14, seconds: 0.163, TFLOP/s/device: 83.184, Tokens/s/device: 12538.341, total_weights: 65536, loss: 6.635, lm_loss: 6.635, perplexity: 761.080 I0422 09:44:35.596947 133766386374464 metric_logger.py:196] completed step: 15, seconds: 0.158, TFLOP/s/device: 85.815, Tokens/s/device: 12935.009, total_weights: 65536, loss: 6.492, lm_loss: 6.492, perplexity: 659.915 I0422 09:44:35.755341 133766386374464 metric_logger.py:196] completed step: 16, seconds: 0.158, TFLOP/s/device: 85.842, Tokens/s/device: 12939.095, total_weights: 65536, loss: 6.380, lm_loss: 6.380, perplexity: 589.948 I0422 09:44:35.913616 133766386374464 metric_logger.py:196] completed step: 17, seconds: 0.158, TFLOP/s/device: 85.765, Tokens/s/device: 12927.416, total_weights: 65536, loss: 6.293, lm_loss: 6.293, perplexity: 540.941 I0422 09:44:36.071913 133766386374464 metric_logger.py:196] completed step: 18, seconds: 0.158, TFLOP/s/device: 85.875, Tokens/s/device: 12944.084, total_weights: 65536, loss: 6.223, lm_loss: 6.223, perplexity: 504.108 I0422 09:44:36.229615 133766386374464 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0422 09:44:36.230599 133766386374464 checkpointing.py:798] Waited 0.0010030269622802734 seconds for step 19 to finish before starting checkpointing. I0422 09:44:36.232732 133766386374464 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 09:44:41.070358 133634308679424 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10/items/array_metadatas/process_5 I0422 09:45:17.654665 133635952862976 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 36.467 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.32058644294739 s) (per-host) I0422 09:45:17.654796 133635952862976 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.258908s. I0422 09:45:26.602943 133635952862976 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 52.207040s. I0422 09:45:26.603238 133633805379328 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 09:45:26.603357 133633805379328 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0422 09:45:26.603406 133633805379328 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0422 09:45:26.604901 133633805379328 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0422 09:45:26.605084 133766386374464 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0422 09:45:26.605205 133766386374464 checkpoint_manager.py:1452] Waiting for previous save to complete took 50.372478 seconds. If this number is high, consider checkpointing less frequently. I0422 09:45:26.606706 133766386374464 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 19 I0422 09:45:26.608960 133766386374464 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19. I0422 09:45:27.325207 133766386374464 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0422 09:45:27.325302 133766386374464 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 I0422 09:45:27.377164 133766386374464 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.052897s I0422 09:45:27.377333 133766386374464 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 26.061 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.059197187423706055 s) (per-host) I0422 09:45:27.377384 133766386374464 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.059259s (batch_requests_ready=0.003236s, total_serialization_initiated=0.055956s, others=0.000067s) I0422 09:45:27.377490 133766386374464 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.063589s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=0.063576) I0422 09:45:27.378181 133766386374464 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19. I0422 09:45:27.378468 133633805379328 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 10:05:27.378443 I0422 09:45:27.380449 133766386374464 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0422 09:45:27.380743 133634308679424 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0422 09:45:27.380907 133766386374464 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776851076.232704, 'wait_for_prev_duration_secs': 50.372477531433105, 'time_between_consecutive_saves_sec': 2.188180446624756, 'checkpointer_blocking_start_time': 1776851126.6067445, 'checkpointer_blocking_duration_secs': 0.7718305587768555, 'get_old_steps_start_time': 1776851127.3785942, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776851076.2308831, 'checkpoint_manager_blocking_duration_secs': 51.149988651275635} I0422 09:45:27.381019 133766386374464 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0422 09:45:27.381061 133766386374464 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 09:45:34.368953 133634862335744 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/linen_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19/items/array_metadatas/process_5 I0422 09:46:12.348906 133633805379328 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 35.082 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.03074550628662 s) (per-host) I0422 09:46:12.349018 133633805379328 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.970469s. I0422 09:46:19.078336 133633805379328 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 51.699771s. I0422 09:46:19.078620 133634308679424 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 09:46:19.078754 133634308679424 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0422 09:46:19.078796 133634308679424 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0422 09:46:19.080226 133634308679424 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0422 09:46:19.080395 133766386374464 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0422 09:46:19.080543 133766386374464 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 09:46:19.081517 133766386374464 metric_logger.py:196] completed step: 19, seconds: 0.158, TFLOP/s/device: 85.785, Tokens/s/device: 12930.517, 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: Wed Apr 22 09:46:32 UTC 2026 EXIT_CODE=0