XPK Start: Sat Apr 18 20:52:29 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-18 20:52:53.515299: 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) I0418 20:52:53.694732 136276464412480 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 20:53:02.735399 136276464412480 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 20:53:02.737863 136276464412480 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-4vs5w-slice-job-0-0.mt-05-fp8-4vs5w:8482 I0418 20:53:04.476060 136276464412480 max_utils.py:284] Jax distributed system initialized! I0418 20:53:09.635074 136276464412480 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 20:53:09.635174 136276464412480 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 20:53:09.635212 136276464412480 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 I0418 20:53:09.635262 136276464412480 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 20:53:10.702991 136276464412480 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 20:53:10.703586 136276464412480 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 20:53:10.703771 136276464412480 checkpointing.py:688] Setting up checkpoint logger... I0418 20:53:10.703818 136276464412480 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 20:53:10.703861 136276464412480 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 20:53:10.704205 136276464412480 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 0x7bf0e07ea600>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 20:53:14.334922 136276464412480 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0418 20:53:14.335109 136276464412480 checkpoint_manager.py:708] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bef92ced730>}, handler_registry=None I0418 20:53:14.335358 136276464412480 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bef92ced730>` 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`. I0418 20:53:14.335409 136276464412480 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bef92d67110>` 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`. I0418 20:53:14.335444 136276464412480 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 0x7bef92ced730>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bef92ced730>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bef92d67110>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bef92d67110>}). I0418 20:53:14.335756 136276464412480 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 20:53:14.335860 136276464412480 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7bdd041d6d40> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 20:53:15.852429 136276464412480 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints I0418 20:53:17.174968 136276464412480 checkpoint_manager.py:929] [process=3][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_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bef92cecdd0> I0418 20:53:17.175139 136276464412480 checkpointing.py:302] Checkpoint manager created! I0418 20:53:18.259085 136276464412480 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 20:53:18.259215 136276464412480 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 20:53:18.646624 136276464412480 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 20:53:18.646713 136276464412480 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 20:53:18.663338 136276464412480 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 20:53:18.663394 136276464412480 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 20:53:18.715563 136276464412480 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 20:53:18.715645 136276464412480 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 20:53:18.732349 136276464412480 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 20:53:18.732409 136276464412480 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 20:53:18.748999 136276464412480 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 20:53:18.749062 136276464412480 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 20:53:18.774119 136276464412480 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 20:53:18.774194 136276464412480 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 20:53:18.825096 136276464412480 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 20:53:18.825178 136276464412480 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 20:53:19.265592 136276464412480 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0418 20:53:19.265704 136276464412480 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 20:53:21.189401 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189528 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189577 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189619 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189656 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189688 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189718 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189748 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189778 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189807 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189836 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189864 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189892 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189919 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.189947 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.189975 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190004 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190032 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190060 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190086 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190119 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190147 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190173 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190201 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190227 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190276 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190305 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190333 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190359 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190386 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190413 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190439 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190465 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190491 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190517 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190544 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190570 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190596 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190623 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190650 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190677 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0418 20:53:21.190703 136276464412480 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0418 20:53:21.190755 136276464412480 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 20:53:21.190809 136276464412480 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 20:53:21.190845 136276464412480 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 20:53:21.190891 136276464412480 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0418 20:53:21.190936 136276464412480 maxtext_utils.py:1827] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 20:53:21.190968 136276464412480 maxtext_utils.py:1827] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 20:53:21.191016 136276464412480 maxtext_utils.py:1827] 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) I0418 20:53:21.191064 136276464412480 maxtext_utils.py:1827] 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') I0418 20:53:21.191105 136276464412480 maxtext_utils.py:1827] 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) I0418 20:53:21.191140 136276464412480 maxtext_utils.py:1827] 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) I0418 20:53:21.191184 136276464412480 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 20:53:21.191227 136276464412480 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 20:53:22.835964 136276464412480 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 20:53:22.836063 136276464412480 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 20:53:22.851321 136276464412480 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 20:53:22.851382 136276464412480 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 20:53:36.469499 136276464412480 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. I0418 20:53:36.470394 136276464412480 metric_logger.py:301] number parameters: 1.105 billion I0418 20:53:51.561783 136276464412480 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0418 20:53:51.730463 136276464412480 checkpointing.py:798] Waited 0.1686546802520752 seconds for step 0 to finish before starting checkpointing. I0418 20:53:51.733145 136276464412480 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 20:53:51.735279 136276464412480 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0418 20:53:51.737106 136276464412480 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0. I0418 20:53:52.872135 136276464412480 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 20:53:52.873037 136276464412480 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0418 20:53:52.873157 136276464412480 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 I0418 20:53:53.277027 136276464412480 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.405020s I0418 20:53:53.277210 136276464412480 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 3.731 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.413482666015625 s) (per-host) I0418 20:53:53.277275 136276464412480 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.413558s (batch_requests_ready=0.003619s, total_serialization_initiated=0.409856s, others=0.000082s) I0418 20:53:53.277378 136276464412480 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.417807s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.417790) I0418 20:53:53.278255 136276464412480 event_tracking.py:125] [process=3] [async] Finished blocking save in 1.54 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0. I0418 20:53:53.278543 136150018799360 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 21:13:53.278509 I0418 20:53:53.295178 136276464412480 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 20:53:53.295514 136149470304000 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0418 20:53:53.295667 136276464412480 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776545631.7331266, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776545631.7353635, 'checkpointer_blocking_duration_secs': 1.5433361530303955, 'get_old_steps_start_time': 1776545633.2787263, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776545631.7310588, 'checkpoint_manager_blocking_duration_secs': 1.564570426940918} I0418 20:53:53.295793 136276464412480 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0418 20:53:53.295846 136276464412480 max_utils.py:750] Memstats: After params initialized: I0418 20:53:53.295903 136276464412480 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_10(process=3,(2,2,0,0)) I0418 20:53:53.295936 136276464412480 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_11(process=3,(3,2,0,0)) I0418 20:53:53.295963 136276464412480 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_14(process=3,(2,3,0,0)) I0418 20:53:53.295990 136276464412480 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_15(process=3,(3,3,0,0)) I0418 20:53:53.610911 136276464412480 metric_logger.py:196] completed step: 0, seconds: 15.091, TFLOP/s/device: 0.900, Tokens/s/device: 135.707, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52776.805 I0418 20:53:53.812427 136276464412480 metric_logger.py:196] completed step: 1, seconds: 2.047, TFLOP/s/device: 6.637, Tokens/s/device: 1000.463, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52761.840 I0418 20:53:54.227113 136276464412480 metric_logger.py:196] completed step: 2, seconds: 0.044, TFLOP/s/device: 309.326, Tokens/s/device: 46624.929, total_weights: 65536, loss: 10.816, lm_loss: 10.816, perplexity: 49832.398 I0418 20:53:54.385426 136276464412480 metric_logger.py:196] completed step: 3, seconds: 0.415, TFLOP/s/device: 32.766, Tokens/s/device: 4938.796, total_weights: 65536, loss: 10.431, lm_loss: 10.431, perplexity: 33901.820 I0418 20:53:54.702967 136276464412480 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.351, Tokens/s/device: 12412.798, total_weights: 65536, loss: 9.992, lm_loss: 9.992, perplexity: 21847.639 I0418 20:53:54.709403 136276464412480 metric_logger.py:196] completed step: 5, seconds: 0.157, TFLOP/s/device: 86.281, Tokens/s/device: 13005.156, total_weights: 65536, loss: 9.549, lm_loss: 9.549, perplexity: 14035.006 I0418 20:53:58.474550 2795 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 20:54:00.429903 136149478696704 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/0/items/array_metadatas/process_3 I0418 20:54:15.371325 136150018799360 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 70.188 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 22.507556438446045 s) (per-host) I0418 20:54:15.371451 136150018799360 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 22.092796s. I0418 20:54:20.453289 136276464412480 metric_logger.py:196] completed step: 6, seconds: 0.318, TFLOP/s/device: 42.704, Tokens/s/device: 6436.831, total_weights: 65536, loss: 9.111, lm_loss: 9.111, perplexity: 9058.688 I0418 20:54:20.611787 136276464412480 metric_logger.py:196] completed step: 7, seconds: 25.586, TFLOP/s/device: 0.531, Tokens/s/device: 80.044, total_weights: 65536, loss: 8.685, lm_loss: 8.685, perplexity: 5914.720 I0418 20:54:20.770311 136276464412480 metric_logger.py:196] completed step: 8, seconds: 0.164, TFLOP/s/device: 82.941, Tokens/s/device: 12501.831, total_weights: 65536, loss: 8.281, lm_loss: 8.281, perplexity: 3950.010 I0418 20:54:20.775351 136276464412480 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0418 20:54:21.087299 136276464412480 checkpointing.py:798] Waited 0.3118858337402344 seconds for step 10 to finish before starting checkpointing. I0418 20:54:21.090123 136276464412480 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 20:54:23.222090 136150018799360 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 29.943418s. I0418 20:54:23.222397 136149470304000 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 20:54:23.222521 136149470304000 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0418 20:54:23.222572 136149470304000 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 20:54:23.224054 136149470304000 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 20:54:23.224174 136276464412480 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 20:54:23.224256 136276464412480 checkpoint_manager.py:1452] Waiting for previous save to complete took 2.134136 seconds. If this number is high, consider checkpointing less frequently. I0418 20:54:23.226313 136276464412480 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 10 I0418 20:54:23.228467 136276464412480 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10. I0418 20:54:23.946530 136276464412480 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0418 20:54:23.946629 136276464412480 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 I0418 20:54:23.999984 136276464412480 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.054457s I0418 20:54:24.000154 136276464412480 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 25.351 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.060855865478515625 s) (per-host) I0418 20:54:24.000204 136276464412480 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.060916s (batch_requests_ready=0.003315s, total_serialization_initiated=0.057533s, others=0.000067s) I0418 20:54:24.000320 136276464412480 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.065513s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.065498) I0418 20:54:24.000996 136276464412480 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10. I0418 20:54:24.001329 136149470304000 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 21:14:24.001293 I0418 20:54:24.003366 136276464412480 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0418 20:54:24.003601 136147913144064 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0418 20:54:24.003722 136276464412480 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776545661.0900865, 'wait_for_prev_duration_secs': 2.1341359615325928, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776545663.226395, 'checkpointer_blocking_duration_secs': 0.7750704288482666, 'get_old_steps_start_time': 1776545664.001487, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776545661.0880241, 'checkpoint_manager_blocking_duration_secs': 2.9156649112701416} I0418 20:54:24.003849 136276464412480 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0418 20:54:24.004481 136276464412480 metric_logger.py:196] completed step: 9, seconds: 0.158, TFLOP/s/device: 85.738, Tokens/s/device: 12923.337, total_weights: 65536, loss: 7.908, lm_loss: 7.908, perplexity: 2720.116 I0418 20:54:24.011792 136276464412480 metric_logger.py:196] completed step: 10, seconds: 0.159, TFLOP/s/device: 85.587, Tokens/s/device: 12900.625, total_weights: 65536, loss: 7.568, lm_loss: 7.568, perplexity: 1936.198 I0418 20:54:24.170892 136276464412480 metric_logger.py:196] completed step: 11, seconds: 3.234, TFLOP/s/device: 4.201, Tokens/s/device: 633.234, total_weights: 65536, loss: 7.288, lm_loss: 7.288, perplexity: 1463.272 I0418 20:54:24.329130 136276464412480 metric_logger.py:196] completed step: 12, seconds: 0.007, TFLOP/s/device: 2052.126, Tokens/s/device: 309318.834, total_weights: 65536, loss: 7.032, lm_loss: 7.032, perplexity: 1132.061 I0418 20:54:24.487495 136276464412480 metric_logger.py:196] completed step: 13, seconds: 0.160, TFLOP/s/device: 85.001, Tokens/s/device: 12812.332, total_weights: 65536, loss: 6.815, lm_loss: 6.815, perplexity: 911.031 I0418 20:54:24.646626 136276464412480 metric_logger.py:196] completed step: 14, seconds: 0.158, TFLOP/s/device: 85.858, Tokens/s/device: 12941.385, total_weights: 65536, loss: 6.635, lm_loss: 6.635, perplexity: 761.080 I0418 20:54:25.207544 136276464412480 metric_logger.py:196] completed step: 15, seconds: 0.158, TFLOP/s/device: 85.739, Tokens/s/device: 12923.582, total_weights: 65536, loss: 6.492, lm_loss: 6.492, perplexity: 659.915 I0418 20:54:25.365741 136276464412480 metric_logger.py:196] completed step: 16, seconds: 0.556, TFLOP/s/device: 24.428, Tokens/s/device: 3682.122, total_weights: 65536, loss: 6.380, lm_loss: 6.380, perplexity: 589.948 I0418 20:54:25.524312 136276464412480 metric_logger.py:196] completed step: 17, seconds: 0.164, TFLOP/s/device: 83.008, Tokens/s/device: 12511.913, total_weights: 65536, loss: 6.293, lm_loss: 6.293, perplexity: 540.941 I0418 20:54:25.682624 136276464412480 metric_logger.py:196] completed step: 18, seconds: 0.158, TFLOP/s/device: 85.830, Tokens/s/device: 12937.215, total_weights: 65536, loss: 6.223, lm_loss: 6.223, perplexity: 504.108 I0418 20:54:25.840430 136276464412480 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0418 20:54:25.841415 136276464412480 checkpointing.py:798] Waited 0.0010058879852294922 seconds for step 19 to finish before starting checkpointing. I0418 20:54:25.843448 136276464412480 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 20:54:31.218118 136145765660416 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/10/items/array_metadatas/process_3 I0418 20:55:07.797439 136149470304000 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 36.020 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.85810208320618 s) (per-host) I0418 20:55:07.797569 136149470304000 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.796129s. I0418 20:55:14.464314 136149470304000 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.462860s. I0418 20:55:14.464599 136147913144064 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 20:55:14.464718 136147913144064 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0418 20:55:14.464770 136147913144064 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0418 20:55:14.466391 136147913144064 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0418 20:55:14.466558 136276464412480 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0418 20:55:14.466681 136276464412480 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.623237 seconds. If this number is high, consider checkpointing less frequently. I0418 20:55:14.468320 136276464412480 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 19 I0418 20:55:14.470536 136276464412480 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19. I0418 20:55:15.192290 136276464412480 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0418 20:55:15.192384 136276464412480 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 I0418 20:55:15.244370 136276464412480 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.053092s I0418 20:55:15.244532 136276464412480 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 25.907 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.059548139572143555 s) (per-host) I0418 20:55:15.244581 136276464412480 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.059606s (batch_requests_ready=0.003252s, total_serialization_initiated=0.056291s, others=0.000063s) I0418 20:55:15.244683 136276464412480 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.063963s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.063951) I0418 20:55:15.245345 136276464412480 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19. I0418 20:55:15.245686 136147913144064 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 21:15:15.245649 I0418 20:55:15.247703 136276464412480 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0418 20:55:15.247997 136145765660416 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0418 20:55:15.248158 136276464412480 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776545665.843419, 'wait_for_prev_duration_secs': 48.623236894607544, 'time_between_consecutive_saves_sec': 2.619321346282959, 'checkpointer_blocking_start_time': 1776545714.4683576, 'checkpointer_blocking_duration_secs': 0.7774693965911865, 'get_old_steps_start_time': 1776545715.2458434, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776545665.8416882, 'checkpoint_manager_blocking_duration_secs': 49.40643858909607} I0418 20:55:15.248306 136276464412480 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0418 20:55:15.248352 136276464412480 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 20:55:22.432861 136146839402240 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 81 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260418_204152_05_fp8/checkpoints/19/items/array_metadatas/process_3 I0418 20:56:00.261205 136147913144064 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 35.046 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.076181411743164 s) (per-host) I0418 20:56:00.261343 136147913144064 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.015544s. I0418 20:56:05.969056 136147913144064 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.723242s. I0418 20:56:05.969380 136145765660416 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 20:56:05.969511 136145765660416 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0418 20:56:05.969561 136145765660416 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0418 20:56:05.970920 136145765660416 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0418 20:56:05.971092 136276464412480 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0418 20:56:05.971252 136276464412480 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 20:56:05.972198 136276464412480 metric_logger.py:196] completed step: 19, seconds: 0.159, TFLOP/s/device: 85.719, Tokens/s/device: 12920.565, 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: Sat Apr 18 20:56:14 UTC 2026 EXIT_CODE=0