XPK Start: Mon Apr 20 15:53:17 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 15:53:41.619756: 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 15:53:41.799999 132223796635456 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 15:53:50.842298 132223796635456 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 15:53:50.844760 132223796635456 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-4kceq-slice-job-0-0.mt-05-fp8-4kceq:8482 I0420 15:53:51.779000 132223796635456 max_utils.py:284] Jax distributed system initialized! I0420 15:53:57.773299 132223796635456 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 15:53:57.773400 132223796635456 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 15:53:57.773442 132223796635456 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 I0420 15:53:57.773478 132223796635456 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 15:53:58.820404 132223796635456 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 15:53:58.821008 132223796635456 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 15:53:58.821203 132223796635456 checkpointing.py:688] Setting up checkpoint logger... I0420 15:53:58.821257 132223796635456 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 15:53:58.821301 132223796635456 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 15:53:58.821626 132223796635456 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 0x7841253766c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 15:54:01.482985 132223796635456 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 15:54:01.483236 132223796635456 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x783ffc7b8b30>}, handler_registry=None I0420 15:54:01.483474 132223796635456 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x783ffc7b8b30>` 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 15:54:01.483521 132223796635456 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x783ffc82fb90>` 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 15:54:01.483557 132223796635456 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 0x783ffc7b8b30>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x783ffc7b8b30>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x783ffc82fb90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x783ffc82fb90>}). I0420 15:54:01.483874 132223796635456 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 15:54:01.483944 132223796635456 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x782bf82d2c00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 15:54:02.599140 132223796635456 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints I0420 15:54:03.030710 132223796635456 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x783ffc82fb30> I0420 15:54:03.030893 132223796635456 checkpointing.py:302] Checkpoint manager created! I0420 15:54:04.107374 132223796635456 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 15:54:04.107482 132223796635456 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 15:54:04.492401 132223796635456 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 15:54:04.492493 132223796635456 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 15:54:04.508905 132223796635456 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 15:54:04.508963 132223796635456 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 15:54:04.561748 132223796635456 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 15:54:04.561832 132223796635456 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 15:54:04.578290 132223796635456 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 15:54:04.578349 132223796635456 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 15:54:04.594792 132223796635456 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 15:54:04.594852 132223796635456 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 15:54:04.619590 132223796635456 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 15:54:04.619658 132223796635456 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 15:54:04.672392 132223796635456 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 15:54:04.672469 132223796635456 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 15:54:05.117708 132223796635456 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 15:54:05.117819 132223796635456 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 15:54:07.037693 132223796635456 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: () I0420 15:54:07.037809 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.037858 132223796635456 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: () I0420 15:54:07.037897 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_0/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.037938 132223796635456 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: () I0420 15:54:07.037974 132223796635456 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: () I0420 15:54:07.038007 132223796635456 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: () I0420 15:54:07.038038 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038069 132223796635456 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: () I0420 15:54:07.038120 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wi_1/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038153 132223796635456 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: () I0420 15:54:07.038185 132223796635456 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: () I0420 15:54:07.038214 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0420 15:54:07.038242 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038270 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_amax_history Shape: float32[16,1024] Logical: P() Physical: () I0420 15:54:07.038300 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038330 132223796635456 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: () I0420 15:54:07.038366 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/mlp/wo/Fp8DirectDotGeneralOp_0/output_grad_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038395 132223796635456 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: () I0420 15:54:07.038423 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038454 132223796635456 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: () I0420 15:54:07.038496 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/key/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038542 132223796635456 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: () I0420 15:54:07.038575 132223796635456 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: () I0420 15:54:07.038604 132223796635456 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: () I0420 15:54:07.038631 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038662 132223796635456 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: () I0420 15:54:07.038691 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/out/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038718 132223796635456 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: () I0420 15:54:07.038745 132223796635456 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: () I0420 15:54:07.038772 132223796635456 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: () I0420 15:54:07.038799 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038826 132223796635456 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: () I0420 15:54:07.038853 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/query/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038880 132223796635456 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: () I0420 15:54:07.038909 132223796635456 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: () I0420 15:54:07.038937 132223796635456 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: () I0420 15:54:07.038964 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/input_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.038991 132223796635456 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: () I0420 15:54:07.039017 132223796635456 maxtext_utils.py:1827] params/_overwrite_with_gradient/decoder/layers/self_attention/value/Fp8DirectDotGeneralOp_0/kernel_scale Shape: float32[16,1] Logical: P() Physical: () I0420 15:54:07.039044 132223796635456 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: () I0420 15:54:07.039071 132223796635456 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: () I0420 15:54:07.039151 132223796635456 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0420 15:54:07.039209 132223796635456 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) I0420 15:54:07.039248 132223796635456 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) I0420 15:54:07.039296 132223796635456 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') I0420 15:54:07.039343 132223796635456 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) I0420 15:54:07.039376 132223796635456 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) I0420 15:54:07.039422 132223796635456 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) I0420 15:54:07.039469 132223796635456 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') I0420 15:54:07.039504 132223796635456 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) I0420 15:54:07.039537 132223796635456 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) I0420 15:54:07.039592 132223796635456 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0420 15:54:07.039635 132223796635456 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0420 15:54:08.685395 132223796635456 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 15:54:08.685488 132223796635456 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 15:54:08.700819 132223796635456 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 15:54:08.700885 132223796635456 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 15:54:22.303114 132223796635456 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 15:54:22.311052 132223796635456 metric_logger.py:301] number parameters: 1.105 billion I0420 15:54:38.148473 132223796635456 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 15:54:38.316880 132223796635456 checkpointing.py:798] Waited 0.16838431358337402 seconds for step 0 to finish before starting checkpointing. I0420 15:54:38.319318 132223796635456 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 15:54:38.321157 132223796635456 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 15:54:38.322524 132223796635456 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0. I0420 15:54:38.638806 132223796635456 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 15:54:38.656179 132223796635456 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0420 15:54:38.656252 132223796635456 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 15:54:39.072210 132223796635456 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.416855s I0420 15:54:39.072873 132223796635456 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 3.614 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.42687153816223145 s) (per-host) I0420 15:54:39.072936 132223796635456 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.426953s (batch_requests_ready=0.003863s, total_serialization_initiated=0.422512s, others=0.000578s) I0420 15:54:39.073016 132223796635456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.430772s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.430753) I0420 15:54:39.073853 132223796635456 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0. I0420 15:54:39.074218 132093369407232 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 16:14:39.074179 I0420 15:54:39.080624 132223796635456 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 15:54:39.080945 132092834858752 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 15:54:39.081112 132223796635456 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776700478.3192992, 'wait_for_prev_duration_secs': 6.008148193359375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776700478.3211966, 'checkpointer_blocking_duration_secs': 0.7531704902648926, 'get_old_steps_start_time': 1776700479.0743926, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776700478.317489, 'checkpoint_manager_blocking_duration_secs': 0.7635750770568848} I0420 15:54:39.081217 132223796635456 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 15:54:39.081269 132223796635456 max_utils.py:750] Memstats: After params initialized: I0420 15:54:39.081320 132223796635456 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_0(process=0,(0,0,0,0)) I0420 15:54:39.081354 132223796635456 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_1(process=0,(1,0,0,0)) I0420 15:54:39.081382 132223796635456 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_4(process=0,(0,1,0,0)) I0420 15:54:39.081409 132223796635456 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_5(process=0,(1,1,0,0)) I0420 15:54:39.328159 132093361014528 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 15:54:39.422617 132223796635456 metric_logger.py:196] completed step: 0, seconds: 15.392, TFLOP/s/device: 0.883, Tokens/s/device: 133.059, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52776.805 I0420 15:54:39.426073 132223796635456 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/tensorboard/' I0420 15:54:39.837856 132223796635456 metric_logger.py:196] completed step: 1, seconds: 1.272, TFLOP/s/device: 10.684, Tokens/s/device: 1610.455, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52761.840 I0420 15:54:39.994561 132223796635456 metric_logger.py:196] completed step: 2, seconds: 0.416, TFLOP/s/device: 32.637, Tokens/s/device: 4919.470, total_weights: 65536, loss: 10.816, lm_loss: 10.816, perplexity: 49832.398 I0420 15:54:40.153071 132223796635456 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 1477.344, Tokens/s/device: 222681.309, total_weights: 65536, loss: 10.431, lm_loss: 10.431, perplexity: 33901.820 I0420 15:54:41.308235 132093344229120 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 15:54:41.907864 132092298721024 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776700481037130553, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776700481037130553, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 15:54:42.516727 2883 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 15:54:44.759535 132092851644160 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items/array_metadatas/process_0 I0420 15:55:02.518937 132092843251456 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 20.011967s (commit=18.870129s, array_metadata_write=1.141838s) I0420 15:55:02.520069 132093369407232 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 66.171 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 23.874037504196167 s) (per-host) I0420 15:55:02.520216 132093369407232 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 23.445884s. I0420 15:55:03.213387 132093369407232 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776700481037130553, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 15:55:04.709429 132223796635456 metric_logger.py:196] completed step: 4, seconds: 0.158, TFLOP/s/device: 86.060, Tokens/s/device: 12971.877, total_weights: 65536, loss: 9.992, lm_loss: 9.992, perplexity: 21847.639 I0420 15:55:04.719442 132223796635456 metric_logger.py:196] completed step: 5, seconds: 0.159, TFLOP/s/device: 85.561, Tokens/s/device: 12896.725, total_weights: 65536, loss: 9.549, lm_loss: 9.549, perplexity: 14035.006 I0420 15:55:04.876434 132223796635456 metric_logger.py:196] completed step: 6, seconds: 24.556, TFLOP/s/device: 0.553, Tokens/s/device: 83.402, total_weights: 65536, loss: 9.111, lm_loss: 9.111, perplexity: 9058.688 I0420 15:55:04.945512 132092298721024 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776700481037130553, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/_CHECKPOINT_METADATA I0420 15:55:05.034548 132223796635456 metric_logger.py:196] completed step: 7, seconds: 0.008, TFLOP/s/device: 1784.962, Tokens/s/device: 269048.870, total_weights: 65536, loss: 8.685, lm_loss: 8.685, perplexity: 5914.720 I0420 15:55:05.192686 132223796635456 metric_logger.py:196] completed step: 8, seconds: 0.158, TFLOP/s/device: 85.860, Tokens/s/device: 12941.712, total_weights: 65536, loss: 8.281, lm_loss: 8.281, perplexity: 3950.010 I0420 15:55:05.199876 132223796635456 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 15:55:05.337622 132093369407232 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000463s. I0420 15:55:05.352490 132093369407232 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 15:55:05.509278 132223796635456 checkpointing.py:798] Waited 0.30936527252197266 seconds for step 10 to finish before starting checkpointing. I0420 15:55:05.511913 132223796635456 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 15:55:06.445386 132093369407232 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 3.102159s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 15:55:06.447299 132093369407232 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0/items I0420 15:55:06.989241 132093369407232 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 15:55:08.440441 132093369407232 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0`. I0420 15:55:08.441298 132093369407232 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 30.12 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/0 I0420 15:55:08.442737 132093369407232 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 29.368402s. I0420 15:55:08.442953 132092834858752 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 15:55:08.443066 132092834858752 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 15:55:08.443149 132092834858752 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 15:55:08.446072 132092834858752 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 15:55:08.446217 132223796635456 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 15:55:08.446328 132223796635456 checkpoint_manager.py:1452] Waiting for previous save to complete took 2.934405 seconds. If this number is high, consider checkpointing less frequently. I0420 15:55:08.448109 132223796635456 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 10 I0420 15:55:08.450112 132223796635456 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10. I0420 15:55:09.182307 132223796635456 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0420 15:55:09.182402 132223796635456 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 15:55:09.223522 132223796635456 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042029s I0420 15:55:09.224382 132223796635456 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 30.817 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.050061941146850586 s) (per-host) I0420 15:55:09.224444 132223796635456 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.050140s (batch_requests_ready=0.003576s, total_serialization_initiated=0.045796s, others=0.000767s) I0420 15:55:09.224523 132223796635456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054128s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.054111) I0420 15:55:09.225264 132223796635456 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10. I0420 15:55:09.225584 132092834858752 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 16:15:09.225548 I0420 15:55:09.227732 132223796635456 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 15:55:09.227996 132091786295040 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 15:55:09.228175 132223796635456 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776700505.5118823, 'wait_for_prev_duration_secs': 2.9344050884246826, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776700508.4481487, 'checkpointer_blocking_duration_secs': 0.7775681018829346, 'get_old_steps_start_time': 1776700509.2257404, 'get_old_steps_duration_secs': 3.266334533691406e-05, 'checkpoint_manager_blocking_start_time': 1776700505.510127, 'checkpoint_manager_blocking_duration_secs': 3.7180120944976807} I0420 15:55:09.228290 132223796635456 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 15:55:09.229014 132223796635456 metric_logger.py:196] completed step: 9, seconds: 0.158, TFLOP/s/device: 86.176, Tokens/s/device: 12989.319, total_weights: 65536, loss: 7.908, lm_loss: 7.908, perplexity: 2720.116 I0420 15:55:09.238985 132223796635456 metric_logger.py:196] completed step: 10, seconds: 0.158, TFLOP/s/device: 85.949, Tokens/s/device: 12955.220, total_weights: 65536, loss: 7.568, lm_loss: 7.568, perplexity: 1936.198 I0420 15:55:09.394609 132223796635456 metric_logger.py:196] completed step: 11, seconds: 4.037, TFLOP/s/device: 3.366, Tokens/s/device: 507.316, total_weights: 65536, loss: 7.288, lm_loss: 7.288, perplexity: 1463.272 I0420 15:55:09.458324 132093369407232 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 15:55:09.552844 132223796635456 metric_logger.py:196] completed step: 12, seconds: 0.008, TFLOP/s/device: 1654.344, Tokens/s/device: 249360.770, total_weights: 65536, loss: 7.032, lm_loss: 7.032, perplexity: 1132.061 I0420 15:55:09.711192 132223796635456 metric_logger.py:196] completed step: 13, seconds: 0.157, TFLOP/s/device: 86.655, Tokens/s/device: 13061.558, total_weights: 65536, loss: 6.815, lm_loss: 6.815, perplexity: 911.031 I0420 15:55:10.217511 132223796635456 metric_logger.py:196] completed step: 14, seconds: 0.158, TFLOP/s/device: 85.921, Tokens/s/device: 12950.960, total_weights: 65536, loss: 6.635, lm_loss: 6.635, perplexity: 761.080 I0420 15:55:10.374563 132223796635456 metric_logger.py:196] completed step: 15, seconds: 0.657, TFLOP/s/device: 20.694, Tokens/s/device: 3119.260, total_weights: 65536, loss: 6.492, lm_loss: 6.492, perplexity: 659.915 I0420 15:55:10.532643 132223796635456 metric_logger.py:196] completed step: 16, seconds: 0.007, TFLOP/s/device: 1922.072, Tokens/s/device: 289715.660, total_weights: 65536, loss: 6.380, lm_loss: 6.380, perplexity: 589.948 I0420 15:55:10.690814 132223796635456 metric_logger.py:196] completed step: 17, seconds: 0.158, TFLOP/s/device: 85.904, Tokens/s/device: 12948.339, total_weights: 65536, loss: 6.293, lm_loss: 6.293, perplexity: 540.941 I0420 15:55:10.849312 132223796635456 metric_logger.py:196] completed step: 18, seconds: 0.158, TFLOP/s/device: 85.849, Tokens/s/device: 12940.158, total_weights: 65536, loss: 6.223, lm_loss: 6.223, perplexity: 504.108 I0420 15:55:11.006620 132223796635456 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 15:55:11.007647 132223796635456 checkpointing.py:798] Waited 0.001049041748046875 seconds for step 19 to finish before starting checkpointing. I0420 15:55:11.009919 132223796635456 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 15:55:11.769059 132092826466048 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 15:55:15.207862 132092273542912 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items/array_metadatas/process_0 I0420 15:55:54.740137 132092265150208 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.788466s (commit=40.503257s, array_metadata_write=1.285209s) I0420 15:55:54.741383 132092834858752 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.669 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.56702923774719 s) (per-host) I0420 15:55:54.741497 132092834858752 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.515804s. I0420 15:55:56.526409 132092834858752 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 15:55:56.676272 132092834858752 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000477s. I0420 15:55:57.190111 132092834858752 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.508306s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 15:55:57.191777 132092834858752 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10/items I0420 15:55:58.160265 132092834858752 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 15:55:59.612737 132092834858752 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10`. I0420 15:55:59.613468 132092834858752 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.17 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/10 I0420 15:55:59.614931 132092834858752 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.389236s. I0420 15:55:59.615175 132091786295040 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 15:55:59.615296 132091786295040 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 15:55:59.615359 132091786295040 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 15:55:59.617147 132091786295040 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 15:55:59.617268 132223796635456 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0420 15:55:59.617329 132223796635456 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.607420 seconds. If this number is high, consider checkpointing less frequently. I0420 15:55:59.619075 132223796635456 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 19 I0420 15:55:59.621174 132223796635456 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19. I0420 15:56:00.349754 132223796635456 jax_array_handlers.py:360] Scheduling D2H of 81 prioritized jax.Array. I0420 15:56:00.349846 132223796635456 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 15:56:00.389056 132223796635456 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040081s I0420 15:56:00.389859 132223796635456 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 32.089 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04807734489440918 s) (per-host) I0420 15:56:00.389927 132223796635456 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048156s (batch_requests_ready=0.003480s, total_serialization_initiated=0.043979s, others=0.000697s) I0420 15:56:00.390009 132223796635456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.052586s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.052573) I0420 15:56:00.390814 132223796635456 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19. I0420 15:56:00.391128 132092273542912 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 16:16:00.391093 I0420 15:56:00.393096 132223796635456 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 15:56:00.393309 132091786295040 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 15:56:00.393430 132223796635456 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776700511.009892, 'wait_for_prev_duration_secs': 48.607420444488525, 'time_between_consecutive_saves_sec': 2.5637686252593994, 'checkpointer_blocking_start_time': 1776700559.6191335, 'checkpointer_blocking_duration_secs': 0.772132158279419, 'get_old_steps_start_time': 1776700560.391282, 'get_old_steps_duration_secs': 2.3365020751953125e-05, 'checkpoint_manager_blocking_start_time': 1776700511.0079327, 'checkpoint_manager_blocking_duration_secs': 49.385467767715454} I0420 15:56:00.393537 132223796635456 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 15:56:00.393577 132223796635456 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 15:56:00.593343 132092834858752 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 15:56:02.404212 132094999127808 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 15:56:05.545561 132093369407232 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items/array_metadatas/process_0 I0420 15:56:44.252108 132092826466048 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.712185s (commit=39.404994s, array_metadata_write=1.307192s) I0420 15:56:44.253440 132092273542912 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 35.976 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.91165852546692 s) (per-host) I0420 15:56:44.253503 132092273542912 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.862261s. I0420 15:56:46.444248 132092273542912 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 15:56:46.587497 132092273542912 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000463s. I0420 15:56:47.102537 132092273542912 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.498636s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 15:56:47.104249 132092273542912 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19/items I0420 15:56:47.644124 132092273542912 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 15:56:49.124278 132092273542912 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19`. I0420 15:56:49.125118 132092273542912 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 49.51 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_05_fp8/checkpoints/19 I0420 15:56:49.126522 132092273542912 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 48.735278s. I0420 15:56:49.126743 132091786295040 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 15:56:49.126854 132091786295040 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 15:56:49.126906 132091786295040 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 15:56:49.128800 132091786295040 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 15:56:49.128959 132223796635456 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0420 15:56:49.129124 132223796635456 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 15:56:49.130196 132223796635456 metric_logger.py:196] completed step: 19, seconds: 0.158, TFLOP/s/device: 85.998, Tokens/s/device: 12962.600, 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: Mon Apr 20 15:56:58 UTC 2026 EXIT_CODE=0