XPK Start: Mon Apr 20 21:03:46 UTC 2026 2026-04-20 21:03:50.540035: 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 00:00:1776719030.552605 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776719030.556341 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776719030.567592 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719030.567614 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719030.567616 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776719030.567618 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-20 21:04:09.881844: 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 21:04:10.411581 137679336871744 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-20 21:04:19,453:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0420 21:04:19.453121 137679336871744 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-20 21:04:19,455:jax._src.distributed:157: Connecting to JAX distributed service on mt-05-fp8-khlai-slice-job-0-0.mt-05-fp8-khlai:8482 I0420 21:04:19.455418 137679336871744 distributed.py:157] Connecting to JAX distributed service on mt-05-fp8-khlai-slice-job-0-0.mt-05-fp8-khlai:8482 I0420 21:04:20.283295 137679336871744 max_utils.py:284] Jax distributed system initialized! I0420 21:04:26.307965 137679336871744 max_utils.py:800] System Information: Jax Version: 0.8.1 I0420 21:04:26.308071 137679336871744 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0420 21:04:26.308111 137679336871744 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0420 21:04:26.308147 137679336871744 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 21:04:26.926275 137679336871744 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 21:04:27.384357 137679336871744 checkpointing.py:688] Setting up checkpoint logger... I0420 21:04:27.384484 137679336871744 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 21:04:27.384528 137679336871744 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0420 21:04:27.384814 137679336871744 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 0x7d37671795e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 21:04:30.777566 137679336871744 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 21:04:30.777933 137679336871744 checkpoint_manager.py:709] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d240026a6f0>}, handler_registry=None I0420 21:04:30.778173 137679336871744 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d240026a6f0>` 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 21:04:30.778225 137679336871744 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d24001bf380>` 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 21:04:30.778263 137679336871744 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 0x7d240026a6f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d240026a6f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d24001bf380>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d24001bf380>}). I0420 21:04:30.778596 137679336871744 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0420 21:04:30.778685 137679336871744 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d23e82f87c0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 21:04:31.485830 137679336871744 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints I0420 21:04:31.870430 137679336871744 checkpoint_manager.py:929] [process=7][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_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d24001bd340> I0420 21:04:31.870604 137679336871744 checkpointing.py:302] Checkpoint manager created! I0420 21:04:33.213851 137679336871744 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 21:04:33.213981 137679336871744 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 21:04:35.470970 137679336871744 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 21:04:35.471095 137679336871744 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:04:35.471149 137679336871744 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 21:04:35.471194 137679336871744 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 21:04:35.471232 137679336871744 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:04:35.471268 137679336871744 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 21:04:35.471306 137679336871744 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:04:35.471349 137679336871744 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 21:04:35.471386 137679336871744 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:04:35.471420 137679336871744 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 21:04:35.471453 137679336871744 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 21:04:35.471486 137679336871744 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 21:04:35.883765 137679336871744 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:04:35.883913 137679336871744 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:04:35.889326 137679336871744 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 21:04:35.889425 137679336871744 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:04:36.044490 137679336871744 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:04:36.044623 137679336871744 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:04:36.193900 137679336871744 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 21:04:36.194031 137679336871744 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 21:04:36.386955 137679336871744 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:04:36.387101 137679336871744 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:04:36.536083 137679336871744 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:04:36.536226 137679336871744 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:04:36.685491 137679336871744 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 21:04:36.685642 137679336871744 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:04:36.846207 137679336871744 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 21:04:36.846355 137679336871744 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 21:04:36.925625 137679336871744 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 21:04:36.925774 137679336871744 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 21:04:50.536579 137679336871744 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 21:04:50.540398 137679336871744 metric_logger.py:289] number parameters: 1.104 billion I0420 21:05:03.412485 137679336871744 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 21:05:03.575854 137679336871744 checkpointing.py:798] Waited 0.16334271430969238 seconds for step 0 to finish before starting checkpointing. I0420 21:05:03.578679 137679336871744 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:05:03.580672 137679336871744 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 0 I0420 21:05:03.583705 137679336871744 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0. I0420 21:05:04.602525 137679336871744 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 21:05:04.603516 137679336871744 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0420 21:05:04.603580 137679336871744 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 21:05:04.967463 137679336871744 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.367435s I0420 21:05:04.967636 137679336871744 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.396 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.6437854766845703 s) (per-host) I0420 21:05:04.967706 137679336871744 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.643866s (batch_requests_ready=0.264601s, total_serialization_initiated=0.379177s, others=0.000087s) I0420 21:05:04.967835 137679336871744 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.648211s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=0.648190) I0420 21:05:04.968833 137560839919360 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 21:05:04.969013 137679336871744 async_checkpointer.py:561] Finished blocking save. Time taken: 1.388274s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0. I0420 21:05:05.421123 137679336871744 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:05:05.421512 137554934339328 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:05:05.421701 137679336871744 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776719103.5786438, 'wait_for_prev_duration_secs': 7.963180541992188e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776719103.5807097, 'checkpointer_blocking_duration_secs': 1.3884153366088867, 'get_old_steps_start_time': 1776719104.9691472, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776719103.5763564, 'checkpoint_manager_blocking_duration_secs': 1.8452825546264648} I0420 21:05:05.421826 137679336871744 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 21:05:05.421908 137679336871744 max_utils.py:750] Memstats: After params initialized: I0420 21:05:05.421963 137679336871744 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_26(process=7,(2,6,0,0)) I0420 21:05:05.421998 137679336871744 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_27(process=7,(3,6,0,0)) I0420 21:05:05.422032 137679336871744 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_30(process=7,(2,7,0,0)) I0420 21:05:05.422057 137679336871744 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_31(process=7,(3,7,0,0)) I0420 21:05:05.692386 137679336871744 metric_logger.py:185] completed step: 0, seconds: 12.870, TFLOP/s/device: 1.056, Tokens/s/device: 159.129, total_weights: 65536, loss: 10.872 I0420 21:05:05.958204 137679336871744 metric_logger.py:185] completed step: 1, seconds: 2.277, TFLOP/s/device: 5.967, Tokens/s/device: 899.457, total_weights: 65536, loss: 10.872 I0420 21:05:06.412071 137679336871744 metric_logger.py:185] completed step: 2, seconds: 0.116, TFLOP/s/device: 116.925, Tokens/s/device: 17624.178, total_weights: 65536, loss: 10.856 I0420 21:05:06.567265 137679336871744 metric_logger.py:185] completed step: 3, seconds: 0.431, TFLOP/s/device: 31.521, Tokens/s/device: 4751.167, total_weights: 65536, loss: 10.824 I0420 21:05:06.878934 137679336871744 metric_logger.py:185] completed step: 4, seconds: 0.183, TFLOP/s/device: 74.088, Tokens/s/device: 11167.336, total_weights: 65536, loss: 10.793 I0420 21:05:06.888787 137679336871744 metric_logger.py:185] completed step: 5, seconds: 0.155, TFLOP/s/device: 87.644, Tokens/s/device: 13210.687, total_weights: 65536, loss: 10.763 I0420 21:05:26.783026 2565 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 21:05:26.958349 137679336871744 metric_logger.py:185] completed step: 6, seconds: 0.312, TFLOP/s/device: 43.509, Tokens/s/device: 6558.112, total_weights: 65536, loss: 10.733 I0420 21:05:27.113382 137679336871744 metric_logger.py:185] completed step: 7, seconds: 19.910, TFLOP/s/device: 0.682, Tokens/s/device: 102.862, total_weights: 65536, loss: 10.706 I0420 21:05:27.268571 137679336871744 metric_logger.py:185] completed step: 8, seconds: 0.169, TFLOP/s/device: 80.595, Tokens/s/device: 12148.102, total_weights: 65536, loss: 10.680 I0420 21:05:27.276852 137679336871744 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 21:05:27.579576 137679336871744 checkpointing.py:798] Waited 0.30269622802734375 seconds for step 10 to finish before starting checkpointing. I0420 21:05:27.581785 137679336871744 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:05:28.753627 137560848312064 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/0/items/array_metadatas/process_7 I0420 21:05:43.222979 137560839919360 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 40.612 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.899088621139526 s) (per-host) I0420 21:05:43.223097 137560839919360 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.254150s. I0420 21:05:51.312429 137560839919360 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 46.343465s. I0420 21:05:51.312667 137554934339328 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:05:51.312741 137554934339328 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:05:51.312794 137554934339328 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:05:51.314606 137554934339328 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 21:05:51.314805 137679336871744 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 21:05:51.314944 137679336871744 checkpoint_manager.py:1458] Waiting for previous save to complete took 23.733165 seconds. If this number is high, consider checkpointing less frequently. I0420 21:05:51.316857 137679336871744 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 10 I0420 21:05:51.320451 137679336871744 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10. I0420 21:05:52.297569 137679336871744 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0420 21:05:52.297688 137679336871744 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 21:05:52.353096 137679336871744 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.058957s I0420 21:05:52.353260 137679336871744 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.726 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.32640743255615234 s) (per-host) I0420 21:05:52.353312 137679336871744 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.326467s (batch_requests_ready=0.257886s, total_serialization_initiated=0.068514s, others=0.000067s) I0420 21:05:52.353450 137679336871744 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.330799s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=0.330785) I0420 21:05:52.354293 137554951124736 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 21:05:52.354452 137679336871744 async_checkpointer.py:561] Finished blocking save. Time taken: 1.037527s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10. I0420 21:05:52.789680 137679336871744 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:05:52.790058 137554934339328 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:05:52.790209 137679336871744 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776719127.5817552, 'wait_for_prev_duration_secs': 23.73316502571106, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776719151.3168955, 'checkpointer_blocking_duration_secs': 1.037663221359253, 'get_old_steps_start_time': 1776719152.3545766, 'get_old_steps_duration_secs': 2.3365020751953125e-05, 'checkpoint_manager_blocking_start_time': 1776719127.580024, 'checkpoint_manager_blocking_duration_secs': 25.21014976501465} I0420 21:05:52.790387 137679336871744 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 21:05:52.790873 137679336871744 metric_logger.py:185] completed step: 9, seconds: 0.155, TFLOP/s/device: 87.776, Tokens/s/device: 13230.487, total_weights: 65536, loss: 10.656 I0420 21:05:52.801311 137679336871744 metric_logger.py:185] completed step: 10, seconds: 0.155, TFLOP/s/device: 87.474, Tokens/s/device: 13185.002, total_weights: 65536, loss: 10.636 I0420 21:05:52.952712 137679336871744 metric_logger.py:185] completed step: 11, seconds: 25.522, TFLOP/s/device: 0.532, Tokens/s/device: 80.244, total_weights: 65536, loss: 10.618 I0420 21:05:53.107904 137679336871744 metric_logger.py:185] completed step: 12, seconds: 0.009, TFLOP/s/device: 1477.183, Tokens/s/device: 222657.099, total_weights: 65536, loss: 10.602 I0420 21:05:53.263008 137679336871744 metric_logger.py:185] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.874, Tokens/s/device: 13396.128, total_weights: 65536, loss: 10.588 I0420 21:05:53.418241 137679336871744 metric_logger.py:185] completed step: 14, seconds: 0.155, TFLOP/s/device: 87.587, Tokens/s/device: 13202.086, total_weights: 65536, loss: 10.577 I0420 21:05:53.573621 137679336871744 metric_logger.py:185] completed step: 15, seconds: 0.155, TFLOP/s/device: 87.598, Tokens/s/device: 13203.703, total_weights: 65536, loss: 10.568 I0420 21:05:53.728909 137679336871744 metric_logger.py:185] completed step: 16, seconds: 0.155, TFLOP/s/device: 87.534, Tokens/s/device: 13194.006, total_weights: 65536, loss: 10.562 I0420 21:05:53.884250 137679336871744 metric_logger.py:185] completed step: 17, seconds: 0.155, TFLOP/s/device: 87.455, Tokens/s/device: 13182.117, total_weights: 65536, loss: 10.556 I0420 21:05:54.039483 137679336871744 metric_logger.py:185] completed step: 18, seconds: 0.155, TFLOP/s/device: 87.536, Tokens/s/device: 13194.431, total_weights: 65536, loss: 10.552 I0420 21:05:54.196755 137679336871744 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 21:05:54.197887 137679336871744 checkpointing.py:798] Waited 0.0011425018310546875 seconds for step 19 to finish before starting checkpointing. I0420 21:05:54.199984 137679336871744 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:05:58.348416 137553910953728 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/10/items/array_metadatas/process_7 I0420 21:06:14.323038 137554951124736 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 70.854 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 22.296131134033203 s) (per-host) I0420 21:06:14.323153 137554951124736 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.968748s. I0420 21:06:22.910753 137554951124736 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 30.556334s. I0420 21:06:22.911052 137554934339328 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:06:22.911175 137554934339328 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:06:22.911220 137554934339328 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:06:22.913024 137554934339328 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 21:06:22.913221 137679336871744 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0420 21:06:22.913343 137679336871744 checkpoint_manager.py:1458] Waiting for previous save to complete took 28.713371 seconds. If this number is high, consider checkpointing less frequently. I0420 21:06:22.914775 137679336871744 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 19 I0420 21:06:22.918254 137679336871744 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19. I0420 21:06:23.519509 137679336871744 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0420 21:06:23.519614 137679336871744 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 21:06:23.576578 137679336871744 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.060404s I0420 21:06:23.576751 137679336871744 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.655 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.33142590522766113 s) (per-host) I0420 21:06:23.576802 137679336871744 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.331486s (batch_requests_ready=0.261789s, total_serialization_initiated=0.069632s, others=0.000065s) I0420 21:06:23.576920 137679336871744 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.335481s (all_items=0.000010s, per_item={'items': '0.00001025'}, temp_paths=0.335470) I0420 21:06:23.577776 137553860597504 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0420 21:06:23.577931 137679336871744 async_checkpointer.py:561] Finished blocking save. Time taken: 0.663090s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19. I0420 21:06:24.011196 137679336871744 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 21:06:24.011551 137554934339328 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0420 21:06:24.011747 137679336871744 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776719154.1999564, 'wait_for_prev_duration_secs': 28.71337103843689, 'time_between_consecutive_saves_sec': 2.885309934616089, 'checkpointer_blocking_start_time': 1776719182.9148142, 'checkpointer_blocking_duration_secs': 0.6632282733917236, 'get_old_steps_start_time': 1776719183.578055, 'get_old_steps_duration_secs': 2.0742416381835938e-05, 'checkpoint_manager_blocking_start_time': 1776719154.1981797, 'checkpoint_manager_blocking_duration_secs': 29.81353449821472} I0420 21:06:24.011867 137679336871744 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 21:06:24.011911 137679336871744 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 21:06:28.814752 137556561737472 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260420_190404/nnx_xpk_feat_nnx_set_defaults_true_20260420_190404_05_fp8/checkpoints/19/items/array_metadatas/process_7 I0420 21:06:44.865278 137553860597504 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 73.070 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.61991024017334 s) (per-host) I0420 21:06:44.865394 137553860597504 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.287508s. I0420 21:06:51.728155 137553860597504 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 28.150254s. I0420 21:06:51.728488 137554934339328 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 21:06:51.728608 137554934339328 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0420 21:06:51.728648 137554934339328 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 21:06:51.730324 137554934339328 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 21:06:51.730512 137679336871744 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0420 21:06:51.730700 137679336871744 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 21:06:51.731381 137679336871744 metric_logger.py:185] completed step: 19, seconds: 0.155, TFLOP/s/device: 87.487, Tokens/s/device: 13187.040, 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: Mon Apr 20 21:07:03 UTC 2026 EXIT_CODE=0