XPK Start: Fri Apr 24 16:40:08 UTC 2026 2026-04-24 16:40:12.825404: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1777048812.838503 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1777048812.842506 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1777048812.853776 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777048812.853795 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777048812.853797 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1777048812.853799 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-24 16:40:32.051900: 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) I0424 16:40:32.578886 134957097375552 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-24 16:40:41,620:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0424 16:40:41.620584 134957097375552 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-24 16:40:41,622:jax._src.distributed:157: Connecting to JAX distributed service on mt-05-fp8-051ef-slice-job-0-0.mt-05-fp8-051ef:8482 I0424 16:40:41.622985 134957097375552 distributed.py:157] Connecting to JAX distributed service on mt-05-fp8-051ef-slice-job-0-0.mt-05-fp8-051ef:8482 I0424 16:40:42.692901 134957097375552 max_utils.py:284] Jax distributed system initialized! I0424 16:40:49.411583 134957097375552 max_utils.py:800] System Information: Jax Version: 0.8.1 I0424 16:40:49.411693 134957097375552 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0424 16:40:49.411735 134957097375552 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0424 16:40:49.411770 134957097375552 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 16:40:50.031873 134957097375552 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 16:40:50.494033 134957097375552 checkpointing.py:688] Setting up checkpoint logger... I0424 16:40:50.494158 134957097375552 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 16:40:50.494205 134957097375552 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0424 16:40:50.494430 134957097375552 base_pytree_checkpoint_handler.py:415] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7abda45a9a90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 16:40:53.378849 134957097375552 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 16:40:53.379227 134957097375552 checkpoint_manager.py:709] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa9602ecb90>}, handler_registry=None I0424 16:40:53.379470 134957097375552 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa9602ecb90>` 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`. I0424 16:40:53.379520 134957097375552 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa9602ed340>` 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`. I0424 16:40:53.379557 134957097375552 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 0x7aa9602ecb90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7aa9602ecb90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa9602ed340>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7aa9602ed340>}). I0424 16:40:53.379903 134957097375552 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0424 16:40:53.379992 134957097375552 async_checkpointer.py:177] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7aa9604247c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0424 16:40:54.105618 134957097375552 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints I0424 16:40:54.455878 134957097375552 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/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7aaa480a6120> I0424 16:40:54.456098 134957097375552 checkpointing.py:302] Checkpoint manager created! I0424 16:40:55.803117 134957097375552 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 16:40:55.803244 134957097375552 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 16:40:58.027782 134957097375552 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 16:40:58.027907 134957097375552 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 16:40:58.027974 134957097375552 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 16:40:58.028021 134957097375552 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 16:40:58.028059 134957097375552 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 16:40:58.028094 134957097375552 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 16:40:58.028131 134957097375552 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 16:40:58.028167 134957097375552 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 16:40:58.028201 134957097375552 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 16:40:58.028234 134957097375552 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 16:40:58.028268 134957097375552 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 16:40:58.028300 134957097375552 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 16:40:58.443041 134957097375552 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 16:40:58.443190 134957097375552 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:40:58.448519 134957097375552 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 16:40:58.448613 134957097375552 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:40:58.596454 134957097375552 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 16:40:58.596586 134957097375552 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:40:58.746667 134957097375552 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0424 16:40:58.746818 134957097375552 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 16:40:58.938261 134957097375552 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 16:40:58.938402 134957097375552 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:40:59.084465 134957097375552 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 16:40:59.084602 134957097375552 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:40:59.227993 134957097375552 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 16:40:59.228153 134957097375552 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:40:59.383970 134957097375552 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0424 16:40:59.384112 134957097375552 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 16:40:59.463303 134957097375552 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 16:40:59.463426 134957097375552 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 16:41:13.021507 134957097375552 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. I0424 16:41:13.025498 134957097375552 metric_logger.py:289] number parameters: 1.104 billion I0424 16:41:26.081942 134957097375552 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 16:41:26.250453 134957097375552 checkpointing.py:798] Waited 0.1685030460357666 seconds for step 0 to finish before starting checkpointing. I0424 16:41:26.252748 134957097375552 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 16:41:26.254514 134957097375552 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 0 I0424 16:41:26.257278 134957097375552 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/0. I0424 16:41:26.883969 134957097375552 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 16:41:26.884982 134957097375552 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0424 16:41:26.885044 134957097375552 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 I0424 16:41:27.243799 134957097375552 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.362327s I0424 16:41:27.243991 134957097375552 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.395 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.644233226776123 s) (per-host) I0424 16:41:27.244046 134957097375552 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.644298s (batch_requests_ready=0.270166s, total_serialization_initiated=0.374059s, others=0.000074s) I0424 16:41:27.244180 134957097375552 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.648397s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.648379) I0424 16:41:27.245159 134830540965632 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 16:41:27.245285 134957097375552 async_checkpointer.py:561] Finished blocking save. Time taken: 0.990701s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/0. I0424 16:41:27.287650 134957097375552 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:41:27.288029 134827848218368 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:41:27.288190 134957097375552 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777048886.2527301, 'wait_for_prev_duration_secs': 6.341934204101562e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777048886.2545526, 'checkpointer_blocking_duration_secs': 0.9909131526947021, 'get_old_steps_start_time': 1777048887.2454865, 'get_old_steps_duration_secs': 2.3603439331054688e-05, 'checkpoint_manager_blocking_start_time': 1777048886.2509604, 'checkpoint_manager_blocking_duration_secs': 1.03719162940979} I0424 16:41:27.288307 134957097375552 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 16:41:27.288388 134957097375552 max_utils.py:750] Memstats: After params initialized: I0424 16:41:27.288441 134957097375552 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_10(process=3,(2,2,0,0)) I0424 16:41:27.288475 134957097375552 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_11(process=3,(3,2,0,0)) I0424 16:41:27.288503 134957097375552 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_14(process=3,(2,3,0,0)) I0424 16:41:27.288528 134957097375552 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_15(process=3,(3,3,0,0)) I0424 16:41:27.561692 134957097375552 metric_logger.py:185] completed step: 0, seconds: 13.054, TFLOP/s/device: 1.041, Tokens/s/device: 156.882, total_weights: 65536, loss: 10.872 I0424 16:41:27.734158 134957097375552 metric_logger.py:185] completed step: 1, seconds: 1.477, TFLOP/s/device: 9.200, Tokens/s/device: 1386.785, total_weights: 65536, loss: 10.872 I0424 16:41:28.223881 134957097375552 metric_logger.py:185] completed step: 2, seconds: 0.022, TFLOP/s/device: 620.445, Tokens/s/device: 93520.252, total_weights: 65536, loss: 10.856 I0424 16:41:28.379080 134957097375552 metric_logger.py:185] completed step: 3, seconds: 0.468, TFLOP/s/device: 29.041, Tokens/s/device: 4377.340, total_weights: 65536, loss: 10.824 I0424 16:41:28.690715 134957097375552 metric_logger.py:185] completed step: 4, seconds: 0.183, TFLOP/s/device: 74.087, Tokens/s/device: 11167.153, total_weights: 65536, loss: 10.793 I0424 16:41:28.700759 134957097375552 metric_logger.py:185] completed step: 5, seconds: 0.155, TFLOP/s/device: 87.724, Tokens/s/device: 13222.714, total_weights: 65536, loss: 10.763 I0424 16:41:48.585620 134957097375552 metric_logger.py:185] completed step: 6, seconds: 0.313, TFLOP/s/device: 43.465, Tokens/s/device: 6551.524, total_weights: 65536, loss: 10.733 I0424 16:41:48.741031 134957097375552 metric_logger.py:185] completed step: 7, seconds: 19.733, TFLOP/s/device: 0.689, Tokens/s/device: 103.786, total_weights: 65536, loss: 10.706 I0424 16:41:48.896075 134957097375552 metric_logger.py:185] completed step: 8, seconds: 0.161, TFLOP/s/device: 84.281, Tokens/s/device: 12703.769, total_weights: 65536, loss: 10.680 I0424 16:41:48.904376 134957097375552 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 16:41:49.207015 134957097375552 checkpointing.py:798] Waited 0.3026103973388672 seconds for step 10 to finish before starting checkpointing. I0424 16:41:49.209248 134957097375552 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:41:49.715153 2639 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 16:41:51.732036 134830549358336 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/0/items/array_metadatas/process_3 I0424 16:42:06.740075 134830540965632 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 39.356 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.14029359817505 s) (per-host) I0424 16:42:06.740176 134830540965632 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.494809s. I0424 16:42:14.242317 134830540965632 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 46.996935s. I0424 16:42:14.242623 134827848218368 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:42:14.242745 134827848218368 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:42:14.242797 134827848218368 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:42:14.244359 134827848218368 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 16:42:14.244527 134957097375552 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0424 16:42:14.244661 134957097375552 checkpoint_manager.py:1458] Waiting for previous save to complete took 25.035419 seconds. If this number is high, consider checkpointing less frequently. I0424 16:42:14.246584 134957097375552 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 10 I0424 16:42:14.250136 134957097375552 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/10. I0424 16:42:14.827093 134957097375552 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0424 16:42:14.827202 134957097375552 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 I0424 16:42:14.879357 134957097375552 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.055730s I0424 16:42:14.879527 134957097375552 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.559 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3383619785308838 s) (per-host) I0424 16:42:14.879577 134957097375552 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.338421s (batch_requests_ready=0.272981s, total_serialization_initiated=0.065373s, others=0.000067s) I0424 16:42:14.879689 134957097375552 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.342571s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.342554) I0424 16:42:14.880557 134827865003776 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 16:42:14.880710 134957097375552 async_checkpointer.py:561] Finished blocking save. Time taken: 0.634055s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/10. I0424 16:42:14.882662 134957097375552 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:42:14.882895 134827848218368 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:42:14.883014 134957097375552 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777048909.2092185, 'wait_for_prev_duration_secs': 25.035419464111328, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777048934.2466223, 'checkpointer_blocking_duration_secs': 0.6342048645019531, 'get_old_steps_start_time': 1777048934.8808458, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1777048909.2074635, 'checkpoint_manager_blocking_duration_secs': 25.67551612854004} I0424 16:42:14.883187 134957097375552 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 16:42:14.883604 134957097375552 metric_logger.py:185] completed step: 9, seconds: 0.155, TFLOP/s/device: 87.492, Tokens/s/device: 13187.719, total_weights: 65536, loss: 10.656 I0424 16:42:14.894745 134957097375552 metric_logger.py:185] completed step: 10, seconds: 0.155, TFLOP/s/device: 87.661, Tokens/s/device: 13213.244, total_weights: 65536, loss: 10.636 I0424 16:42:15.045643 134957097375552 metric_logger.py:185] completed step: 11, seconds: 25.988, TFLOP/s/device: 0.523, Tokens/s/device: 78.807, total_weights: 65536, loss: 10.618 I0424 16:42:15.200770 134957097375552 metric_logger.py:185] completed step: 12, seconds: 0.010, TFLOP/s/device: 1375.494, Tokens/s/device: 207329.419, total_weights: 65536, loss: 10.602 I0424 16:42:15.355885 134957097375552 metric_logger.py:185] completed step: 13, seconds: 0.152, TFLOP/s/device: 89.190, Tokens/s/device: 13443.701, total_weights: 65536, loss: 10.588 I0424 16:42:15.511001 134957097375552 metric_logger.py:185] completed step: 14, seconds: 0.155, TFLOP/s/device: 87.605, Tokens/s/device: 13204.725, total_weights: 65536, loss: 10.577 I0424 16:42:15.666259 134957097375552 metric_logger.py:185] completed step: 15, seconds: 0.155, TFLOP/s/device: 87.658, Tokens/s/device: 13212.733, total_weights: 65536, loss: 10.568 I0424 16:42:15.821284 134957097375552 metric_logger.py:185] completed step: 16, seconds: 0.155, TFLOP/s/device: 87.572, Tokens/s/device: 13199.874, total_weights: 65536, loss: 10.562 I0424 16:42:15.976617 134957097375552 metric_logger.py:185] completed step: 17, seconds: 0.155, TFLOP/s/device: 87.566, Tokens/s/device: 13198.938, total_weights: 65536, loss: 10.556 I0424 16:42:16.131488 134957097375552 metric_logger.py:185] completed step: 18, seconds: 0.155, TFLOP/s/device: 87.610, Tokens/s/device: 13205.491, total_weights: 65536, loss: 10.552 I0424 16:42:16.288892 134957097375552 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 16:42:16.290087 134957097375552 checkpointing.py:798] Waited 0.0012080669403076172 seconds for step 19 to finish before starting checkpointing. I0424 16:42:16.292204 134957097375552 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:42:19.625111 134827319740160 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/10/items/array_metadatas/process_3 I0424 16:42:35.522391 134827865003776 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 75.294 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 20.981175661087036 s) (per-host) I0424 16:42:35.522497 134827865003776 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.641833s. I0424 16:42:43.178556 134827865003776 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 28.297877s. I0424 16:42:43.178818 134827848218368 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:42:43.178875 134827848218368 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:42:43.178921 134827848218368 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:42:43.180726 134827848218368 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 16:42:43.180897 134957097375552 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0424 16:42:43.181035 134957097375552 checkpoint_manager.py:1458] Waiting for previous save to complete took 26.888843 seconds. If this number is high, consider checkpointing less frequently. I0424 16:42:43.182435 134957097375552 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 19 I0424 16:42:43.185838 134957097375552 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/19. I0424 16:42:43.937092 134957097375552 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0424 16:42:43.937196 134957097375552 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 I0424 16:42:43.991188 134957097375552 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.057505s I0424 16:42:43.991350 134957097375552 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.808 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3208916187286377 s) (per-host) I0424 16:42:43.991397 134957097375552 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.320950s (batch_requests_ready=0.254096s, total_serialization_initiated=0.066791s, others=0.000063s) I0424 16:42:43.991512 134957097375552 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.325238s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.325226) I0424 16:42:43.992404 134827311347456 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started. I0424 16:42:43.992570 134957097375552 async_checkpointer.py:561] Finished blocking save. Time taken: 0.810066s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/19. I0424 16:42:44.013914 134957097375552 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 16:42:44.014218 134827848218368 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 16:42:44.014392 134957097375552 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777048936.2921755, 'wait_for_prev_duration_secs': 26.888843059539795, 'time_between_consecutive_saves_sec': 2.047775983810425, 'checkpointer_blocking_start_time': 1777048963.1824737, 'checkpointer_blocking_duration_secs': 0.8102059364318848, 'get_old_steps_start_time': 1777048963.9926927, 'get_old_steps_duration_secs': 2.1219253540039062e-05, 'checkpoint_manager_blocking_start_time': 1777048936.290395, 'checkpoint_manager_blocking_duration_secs': 27.723963975906372} I0424 16:42:44.014532 134957097375552 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 16:42:44.014575 134957097375552 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 16:42:49.473953 134827302954752 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260424_145744/nnx_xpk_feat_nnx_set_defaults_true_20260424_145744_05_fp8/checkpoints/19/items/array_metadatas/process_3 I0424 16:43:05.567457 134827311347456 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 72.145 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.8969566822052 s) (per-host) I0424 16:43:05.567599 134827311347456 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.575076s. I0424 16:43:10.936113 134827311347456 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 26.943575s. I0424 16:43:10.936391 134827848218368 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 16:43:10.936513 134827848218368 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 16:43:10.936554 134827848218368 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 16:43:10.937899 134827848218368 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 16:43:10.938098 134957097375552 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0424 16:43:10.938280 134957097375552 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 16:43:10.938939 134957097375552 metric_logger.py:185] completed step: 19, seconds: 0.155, TFLOP/s/device: 87.530, Tokens/s/device: 13193.411, total_weights: 65536, loss: 10.549 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 16:43:24 UTC 2026 EXIT_CODE=0