XPK Start: Mon Apr 20 19:27:58 UTC 2026 2026-04-20 19:28:02.060414: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776713282.073460 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776713282.077150 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776713282.088662 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776713282.088679 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776713282.088682 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776713282.088684 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 19:28:21.290141: 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) I0420 19:28:21.815165 138380438701888 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 19:28:30,855:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 19:28:30.855206 138380438701888 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 19:28:30,857:jax._src.distributed:157: Connecting to JAX distributed service on mt-05-fp8-yrofe-slice-job-0-0.mt-05-fp8-yrofe:8482 I0420 19:28:30.857467 138380438701888 distributed.py:157] Connecting to JAX distributed service on mt-05-fp8-yrofe-slice-job-0-0.mt-05-fp8-yrofe:8482 I0420 19:28:31.992455 138380438701888 max_utils.py:284] Jax distributed system initialized! I0420 19:28:37.966062 138380438701888 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 19:28:37.966188 138380438701888 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 19:28:37.966230 138380438701888 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0420 19:28:37.966266 138380438701888 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 19:28:38.572387 138380438701888 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 19:28:38.572975 138380438701888 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 19:28:38.573176 138380438701888 checkpointing.py:688] Setting up checkpoint logger... I0420 19:28:38.573228 138380438701888 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 19:28:38.573271 138380438701888 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 19:28:38.573607 138380438701888 base_pytree_checkpoint_handler.py:415] 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 0x7ddab5975310>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 19:28:41.487974 138380438701888 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 19:28:41.488332 138380438701888 checkpoint_manager.py:709] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc6c8212150>}, handler_registry=None I0420 19:28:41.488564 138380438701888 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc6c8212150>` 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`. I0420 19:28:41.488614 138380438701888 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc6c8217260>` 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`. I0420 19:28:41.488658 138380438701888 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 0x7dc6c8212150>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dc6c8212150>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc6c8217260>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc6c8217260>}). I0420 19:28:41.489176 138380438701888 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 19:28:41.489256 138380438701888 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7dc63836c180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 19:28:42.606177 138380438701888 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints I0420 19:28:42.637401 138380438701888 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_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dc6386a5250> I0420 19:28:42.637543 138380438701888 checkpointing.py:302] Checkpoint manager created! I0420 19:28:43.709614 138380438701888 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 19:28:43.709722 138380438701888 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:28:44.080359 138380438701888 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 19:28:44.080457 138380438701888 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:28:44.096053 138380438701888 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 19:28:44.096122 138380438701888 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 19:28:44.155423 138380438701888 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:28:44.155508 138380438701888 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:28:44.171253 138380438701888 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:28:44.171312 138380438701888 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:28:44.186954 138380438701888 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 19:28:44.187015 138380438701888 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:28:44.210565 138380438701888 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 19:28:44.210642 138380438701888 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 19:28:44.269969 138380438701888 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 19:28:44.270051 138380438701888 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 19:28:44.729622 138380438701888 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 19:28:44.729736 138380438701888 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 19:28:46.766681 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.766805 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.766856 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.766896 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.766933 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.766968 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.766999 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767030 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767062 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767178 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767232 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767277 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767319 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767359 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767398 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767437 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767476 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767506 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767553 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767607 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767680 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767724 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767769 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767812 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767857 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767898 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.767942 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.767987 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768027 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768069 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768127 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768171 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768213 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768255 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768294 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768331 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768375 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768419 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768460 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768502 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768540 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: PartitionSpec() Physical: () I0420 19:28:46.768581 138380438701888 maxtext_utils.py:1740] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: PartitionSpec() Physical: () I0420 19:28:46.768654 138380438701888 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0420 19:28:46.768733 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 19:28:46.768792 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 19:28:46.768861 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0420 19:28:46.768931 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0420 19:28:46.768976 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0420 19:28:46.769051 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0420 19:28:46.769140 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0420 19:28:46.769201 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0420 19:28:46.769255 138380438701888 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0420 19:28:46.769322 138380438701888 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0420 19:28:46.769394 138380438701888 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0420 19:28:48.534056 138380438701888 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 19:28:48.534171 138380438701888 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 19:28:48.550075 138380438701888 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 19:28:48.550146 138380438701888 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 19:29:01.683795 138380438701888 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. I0420 19:29:01.690133 138380438701888 metric_logger.py:289] number parameters: 1.105 billion I0420 19:29:16.886421 138380438701888 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 19:29:17.053754 138380438701888 checkpointing.py:798] Waited 0.16731810569763184 seconds for step 0 to finish before starting checkpointing. I0420 19:29:17.056452 138380438701888 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:29:17.058511 138380438701888 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 0 I0420 19:29:17.061573 138380438701888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0. I0420 19:29:17.382529 138380438701888 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 19:29:17.628942 138254410028800 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0 I0420 19:29:17.648361 138380438701888 jax_array_handlers.py:358] Scheduling D2H of 81 prioritized jax.Array. I0420 19:29:17.648460 138380438701888 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 I0420 19:29:18.008801 138380438701888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.361773s I0420 19:29:18.009662 138380438701888 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.490 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.6196041107177734 s) (per-host) I0420 19:29:18.009740 138380438701888 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.619704s (batch_requests_ready=0.251233s, total_serialization_initiated=0.367685s, others=0.000786s) I0420 19:29:18.009823 138380438701888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.623558s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.623539) I0420 19:29:18.010790 138256574297856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 19:29:18.010953 138380438701888 async_checkpointer.py:561] Finished blocking save. Time taken: 0.952371s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0. I0420 19:29:18.123315 138380438701888 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:29:18.123687 138253923514112 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:29:18.123852 138380438701888 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776713357.0564344, 'wait_for_prev_duration_secs': 6.413459777832031e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776713357.0585504, 'checkpointer_blocking_duration_secs': 0.9525091648101807, 'get_old_steps_start_time': 1776713358.0110903, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776713357.0543964, 'checkpoint_manager_blocking_duration_secs': 1.069415807723999} I0420 19:29:18.123969 138380438701888 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 19:29:18.124024 138380438701888 max_utils.py:750] Memstats: After params initialized: I0420 19:29:18.124095 138380438701888 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_0(process=0,(0,0,0,0)) I0420 19:29:18.124132 138380438701888 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_1(process=0,(1,0,0,0)) I0420 19:29:18.124161 138380438701888 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_4(process=0,(0,1,0,0)) I0420 19:29:18.124187 138380438701888 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_5(process=0,(1,1,0,0)) I0420 19:29:18.431839 138380438701888 metric_logger.py:185] completed step: 0, seconds: 14.839, TFLOP/s/device: 0.916, Tokens/s/device: 138.018, total_weights: 65536, loss: 10.874 I0420 19:29:18.434742 138380438701888 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/tensorboard/' I0420 19:29:18.834668 138380438701888 metric_logger.py:185] completed step: 1, seconds: 1.543, TFLOP/s/device: 8.805, Tokens/s/device: 1327.206, total_weights: 65536, loss: 10.874 I0420 19:29:18.991215 138380438701888 metric_logger.py:185] completed step: 2, seconds: 0.404, TFLOP/s/device: 33.623, Tokens/s/device: 5068.078, total_weights: 65536, loss: 10.816 I0420 19:29:19.149526 138380438701888 metric_logger.py:185] completed step: 3, seconds: 0.009, TFLOP/s/device: 1537.527, Tokens/s/device: 231752.857, total_weights: 65536, loss: 10.431 I0420 19:29:39.753153 138256029005568 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776713359309240490, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776713359309240490, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 19:29:39.753555 138256607868672 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/items I0420 19:29:39.755133 138380438701888 metric_logger.py:185] completed step: 4, seconds: 0.157, TFLOP/s/device: 86.329, Tokens/s/device: 13012.428, total_weights: 65536, loss: 9.992 I0420 19:29:39.764404 138380438701888 metric_logger.py:185] completed step: 5, seconds: 0.158, TFLOP/s/device: 85.786, Tokens/s/device: 12930.599, total_weights: 65536, loss: 9.549 I0420 19:29:39.920715 138380438701888 metric_logger.py:185] completed step: 6, seconds: 20.605, TFLOP/s/device: 0.659, Tokens/s/device: 99.393, total_weights: 65536, loss: 9.111 I0420 19:29:40.079157 138380438701888 metric_logger.py:185] completed step: 7, seconds: 0.007, TFLOP/s/device: 1970.863, Tokens/s/device: 297069.916, total_weights: 65536, loss: 8.685 I0420 19:29:40.237148 138380438701888 metric_logger.py:185] completed step: 8, seconds: 0.158, TFLOP/s/device: 86.208, Tokens/s/device: 12994.182, total_weights: 65536, loss: 8.281 I0420 19:29:40.244179 138380438701888 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 19:29:40.553311 138380438701888 checkpointing.py:798] Waited 0.3090989589691162 seconds for step 10 to finish before starting checkpointing. I0420 19:29:40.555967 138380438701888 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:29:41.005219 2648 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 19:29:42.655302 138256591083264 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/items/array_metadatas/process_0 I0420 19:30:00.479247 138256582690560 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 19.480317s (commit=18.226357s, array_metadata_write=1.253960s) I0420 19:30:00.480363 138256574297856 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 36.662 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.090264320373535 s) (per-host) I0420 19:30:00.480506 138256574297856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.469600s. I0420 19:30:01.145565 138256574297856 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776713359309240490, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 19:30:01.898004 138256029005568 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776713359309240490, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 19:30:02.170850 138256574297856 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000465s. I0420 19:30:02.525546 138256574297856 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 19:30:03.450277 138256574297856 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.181217s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/items I0420 19:30:03.452023 138256574297856 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/items I0420 19:30:03.891466 138256574297856 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0 I0420 19:30:05.198067 138256574297856 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0`. I0420 19:30:05.198921 138256574297856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.140345s. directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0 I0420 19:30:05.200390 138256574297856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 47.189483s. I0420 19:30:05.200560 138253923514112 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:30:05.200637 138253923514112 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:30:05.200702 138253923514112 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:30:05.202694 138253923514112 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 19:30:05.202871 138380438701888 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 19:30:05.203007 138380438701888 checkpoint_manager.py:1458] Waiting for previous save to complete took 24.647046 seconds. If this number is high, consider checkpointing less frequently. I0420 19:30:05.204926 138380438701888 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 10 I0420 19:30:05.208505 138380438701888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10. I0420 19:30:05.772561 138256574297856 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10 I0420 19:30:05.781216 138380438701888 jax_array_handlers.py:358] Scheduling D2H of 81 prioritized jax.Array. I0420 19:30:05.781318 138380438701888 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 I0420 19:30:05.821858 138380438701888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041629s I0420 19:30:05.822703 138380438701888 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.911 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.31410884857177734 s) (per-host) I0420 19:30:05.822770 138380438701888 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.314191s (batch_requests_ready=0.267622s, total_serialization_initiated=0.045815s, others=0.000754s) I0420 19:30:05.822849 138380438701888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.318208s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.318192) I0420 19:30:05.823642 138252823045888 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 19:30:05.823748 138380438701888 async_checkpointer.py:561] Finished blocking save. Time taken: 0.618754s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10. I0420 19:30:05.849351 138380438701888 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:30:05.849596 138253923514112 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:30:05.849746 138380438701888 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776713380.5559366, 'wait_for_prev_duration_secs': 24.647045850753784, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776713405.2049649, 'checkpointer_blocking_duration_secs': 0.6188921928405762, 'get_old_steps_start_time': 1776713405.823876, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776713380.5540748, 'checkpoint_manager_blocking_duration_secs': 25.295637845993042} I0420 19:30:05.849854 138380438701888 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 19:30:05.850282 138380438701888 metric_logger.py:185] completed step: 9, seconds: 0.158, TFLOP/s/device: 85.769, Tokens/s/device: 12928.069, total_weights: 65536, loss: 7.908 I0420 19:30:05.859855 138380438701888 metric_logger.py:185] completed step: 10, seconds: 0.158, TFLOP/s/device: 86.134, Tokens/s/device: 12983.061, total_weights: 65536, loss: 7.568 I0420 19:30:06.015286 138380438701888 metric_logger.py:185] completed step: 11, seconds: 25.614, TFLOP/s/device: 0.530, Tokens/s/device: 79.957, total_weights: 65536, loss: 7.288 I0420 19:30:06.173650 138380438701888 metric_logger.py:185] completed step: 12, seconds: 0.008, TFLOP/s/device: 1684.912, Tokens/s/device: 253968.254, total_weights: 65536, loss: 7.032 I0420 19:30:06.331752 138380438701888 metric_logger.py:185] completed step: 13, seconds: 0.156, TFLOP/s/device: 86.856, Tokens/s/device: 13091.867, total_weights: 65536, loss: 6.815 I0420 19:30:06.489755 138380438701888 metric_logger.py:185] completed step: 14, seconds: 0.158, TFLOP/s/device: 85.842, Tokens/s/device: 12939.014, total_weights: 65536, loss: 6.635 I0420 19:30:06.647903 138380438701888 metric_logger.py:185] completed step: 15, seconds: 0.158, TFLOP/s/device: 85.898, Tokens/s/device: 12947.521, total_weights: 65536, loss: 6.492 I0420 19:30:06.805951 138380438701888 metric_logger.py:185] completed step: 16, seconds: 0.158, TFLOP/s/device: 85.940, Tokens/s/device: 12953.827, total_weights: 65536, loss: 6.380 I0420 19:30:06.963962 138380438701888 metric_logger.py:185] completed step: 17, seconds: 0.158, TFLOP/s/device: 86.065, Tokens/s/device: 12972.699, total_weights: 65536, loss: 6.293 I0420 19:30:07.122145 138380438701888 metric_logger.py:185] completed step: 18, seconds: 0.158, TFLOP/s/device: 85.956, Tokens/s/device: 12956.203, total_weights: 65536, loss: 6.223 I0420 19:30:07.279933 138380438701888 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 19:30:07.280915 138380438701888 checkpointing.py:798] Waited 0.0010004043579101562 seconds for step 19 to finish before starting checkpointing. I0420 19:30:07.282917 138380438701888 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:30:07.853197 138254410028800 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10/items I0420 19:30:10.635600 138253897737984 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10/items/array_metadatas/process_0 I0420 19:30:51.210331 138253880952576 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.230465s (commit=40.418787s, array_metadata_write=1.811678s) I0420 19:30:51.212889 138252823045888 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 34.565 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.70429348945618 s) (per-host) I0420 19:30:51.212953 138252823045888 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.389246s. I0420 19:30:52.706233 138252823045888 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000433s. I0420 19:30:52.978574 138252823045888 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 19:30:53.679553 138252823045888 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.571298s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10/items I0420 19:30:53.681278 138252823045888 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10/items I0420 19:30:54.121247 138252823045888 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10 I0420 19:30:55.449064 138252823045888 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10`. I0420 19:30:55.449931 138252823045888 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 50.244939s. directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10 I0420 19:30:55.451344 138252823045888 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 49.627635s. I0420 19:30:55.451483 138253923514112 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:30:55.451547 138253923514112 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:30:55.451605 138253923514112 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:30:55.455119 138253923514112 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 19:30:55.455239 138380438701888 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0420 19:30:55.455306 138380438701888 checkpoint_manager.py:1458] Waiting for previous save to complete took 48.172406 seconds. If this number is high, consider checkpointing less frequently. I0420 19:30:55.456902 138380438701888 checkpoint_manager.py:1518] [process=0] Saving checkpoint at step 19 I0420 19:30:55.460734 138380438701888 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19. I0420 19:30:55.997703 138252823045888 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19 I0420 19:30:56.009974 138380438701888 jax_array_handlers.py:358] Scheduling D2H of 81 prioritized jax.Array. I0420 19:30:56.010092 138380438701888 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 I0420 19:30:56.051626 138380438701888 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042736s I0420 19:30:56.052560 138380438701888 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.124 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3010730743408203 s) (per-host) I0420 19:30:56.052632 138380438701888 base_pytree_checkpoint_handler.py:737] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.301164s (batch_requests_ready=0.253658s, total_serialization_initiated=0.046656s, others=0.000850s) I0420 19:30:56.052721 138380438701888 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.305194s (all_items=0.000010s, per_item={'items': '0.00000954'}, temp_paths=0.305185) I0420 19:30:56.053590 138254424569600 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 19:30:56.053770 138380438701888 async_checkpointer.py:561] Finished blocking save. Time taken: 0.596802s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19. I0420 19:30:56.080899 138380438701888 checkpoint_manager.py:1566] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 19:30:56.081216 138253923514112 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 19:30:56.081378 138380438701888 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776713407.2828884, 'wait_for_prev_duration_secs': 48.17240619659424, 'time_between_consecutive_saves_sec': 2.080153226852417, 'checkpointer_blocking_start_time': 1776713455.4569418, 'checkpointer_blocking_duration_secs': 0.5969367027282715, 'get_old_steps_start_time': 1776713456.0538917, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776713407.2811975, 'checkpoint_manager_blocking_duration_secs': 48.80014944076538} I0420 19:30:56.081482 138380438701888 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 19:30:56.081527 138380438701888 checkpoint_manager.py:2024] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 19:30:57.776587 138254955292416 atomicity.py:137] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19/items I0420 19:31:00.535311 138254441355008 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19/items/array_metadatas/process_0 I0420 19:31:39.557974 138254432962304 base_pytree_checkpoint_handler.py:1232] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.651871s (commit=39.517236s, array_metadata_write=1.134635s) I0420 19:31:39.560032 138254424569600 base_pytree_checkpoint_handler.py:129] [process=0] /jax/checkpoint/write/gbytes_per_sec: 36.061 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.8085150718689 s) (per-host) I0420 19:31:39.560173 138254424569600 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.506447s. I0420 19:31:41.081663 138254424569600 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000416s. I0420 19:31:41.266017 138254424569600 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 19:31:41.895797 138254424569600 base_pytree_checkpoint_handler.py:1356] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.460207s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19/items I0420 19:31:41.897639 138254424569600 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19/items I0420 19:31:42.313014 138254424569600 atomicity.py:608] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19 I0420 19:31:43.725455 138254424569600 atomicity.py:790] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19`. I0420 19:31:43.726241 138254424569600 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.269274s. directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/linen_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19 I0420 19:31:43.727515 138254424569600 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 47.673787s. I0420 19:31:43.727669 138253923514112 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 19:31:43.727737 138253923514112 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 19:31:43.727782 138253923514112 checkpoint_manager.py:2133] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 19:31:43.729742 138253923514112 checkpoint_manager.py:2142] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 19:31:43.729904 138380438701888 checkpoint_manager.py:2036] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0420 19:31:43.730045 138380438701888 checkpoint_manager.py:2013] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 19:31:43.730677 138380438701888 metric_logger.py:185] completed step: 19, seconds: 0.158, TFLOP/s/device: 85.978, Tokens/s/device: 12959.483, total_weights: 65536, loss: 6.167 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 19:31:58 UTC 2026 EXIT_CODE=0