XPK Start: Thu Apr 23 17:41:58 UTC 2026 2026-04-23 17:42:02.665704: 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 00:00:1776966122.679129 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776966122.683179 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776966122.694971 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776966122.694996 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776966122.694998 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776966122.695000 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-23 17:42:21.810701: 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) I0423 17:42:22.338053 139280796612416 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-23 17:42:31,379:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0423 17:42:31.379427 139280796612416 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-23 17:42:31,381:jax._src.distributed:157: Connecting to JAX distributed service on mt-05-fp8-z0e2w-slice-job-0-0.mt-05-fp8-z0e2w:8482 I0423 17:42:31.381835 139280796612416 distributed.py:157] Connecting to JAX distributed service on mt-05-fp8-z0e2w-slice-job-0-0.mt-05-fp8-z0e2w:8482 I0423 17:42:32.614375 139280796612416 max_utils.py:284] Jax distributed system initialized! I0423 17:42:38.601087 139280796612416 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 17:42:38.601214 139280796612416 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 17:42:38.601254 139280796612416 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 I0423 17:42:38.601288 139280796612416 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 17:42:39.225582 139280796612416 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 17:42:39.678728 139280796612416 checkpointing.py:688] Setting up checkpoint logger... I0423 17:42:39.678852 139280796612416 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 17:42:39.678896 139280796612416 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0423 17:42:39.679133 139280796612416 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 0x7eac52175430>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 17:42:43.327388 139280796612416 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 17:42:43.327761 139280796612416 checkpoint_manager.py:709] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e999face0f0>}, handler_registry=None I0423 17:42:43.328024 139280796612416 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e999face0f0>` 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`. I0423 17:42:43.328077 139280796612416 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e999e137e60>` 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`. I0423 17:42:43.328130 139280796612416 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 0x7e999face0f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e999face0f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e999e137e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e999e137e60>}). I0423 17:42:43.328510 139280796612416 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0423 17:42:43.328616 139280796612416 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e999f5b6200> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 17:42:44.043982 139280796612416 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints I0423 17:42:44.046620 139280796612416 checkpoint_manager.py:929] [process=6][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_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e999e1373b0> I0423 17:42:44.046736 139280796612416 checkpointing.py:302] Checkpoint manager created! I0423 17:42:45.375404 139280796612416 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 17:42:45.375533 139280796612416 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0423 17:42:47.607381 139280796612416 maxtext_utils.py:1749] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 17:42:47.607523 139280796612416 maxtext_utils.py:1749] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:42:47.607586 139280796612416 maxtext_utils.py:1749] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 17:42:47.607636 139280796612416 maxtext_utils.py:1749] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 17:42:47.607677 139280796612416 maxtext_utils.py:1749] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:42:47.607715 139280796612416 maxtext_utils.py:1749] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 17:42:47.607753 139280796612416 maxtext_utils.py:1749] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:42:47.607793 139280796612416 maxtext_utils.py:1749] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 17:42:47.607830 139280796612416 maxtext_utils.py:1749] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:42:47.607864 139280796612416 maxtext_utils.py:1749] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 17:42:47.607900 139280796612416 maxtext_utils.py:1749] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 17:42:47.607938 139280796612416 maxtext_utils.py:1749] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 17:42:48.022322 139280796612416 nnx_decoders.py:616] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:42:48.022468 139280796612416 nnx_decoders.py:616] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:42:48.027785 139280796612416 nnx_decoders.py:616] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 17:42:48.027882 139280796612416 nnx_decoders.py:616] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:42:48.176116 139280796612416 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:42:48.176260 139280796612416 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:42:48.325540 139280796612416 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 17:42:48.325674 139280796612416 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 17:42:48.515360 139280796612416 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:42:48.515499 139280796612416 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:42:48.659197 139280796612416 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:42:48.659333 139280796612416 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:42:48.802001 139280796612416 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 17:42:48.802148 139280796612416 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:42:48.955601 139280796612416 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 17:42:48.955741 139280796612416 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 17:42:49.033893 139280796612416 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 17:42:49.034014 139280796612416 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 17:43:02.753659 139280796612416 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. I0423 17:43:02.757581 139280796612416 metric_logger.py:289] number parameters: 1.104 billion I0423 17:43:16.094494 139280796612416 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0423 17:43:16.255604 139280796612416 checkpointing.py:798] Waited 0.1610887050628662 seconds for step 0 to finish before starting checkpointing. I0423 17:43:16.257886 139280796612416 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:43:16.260126 139280796612416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0423 17:43:16.262825 139280796612416 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/0. I0423 17:43:16.868944 139280796612416 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 17:43:16.869979 139280796612416 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0423 17:43:16.870044 139280796612416 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 I0423 17:43:17.238745 139280796612416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.372318s I0423 17:43:17.238919 139280796612416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.412 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.6394765377044678 s) (per-host) I0423 17:43:17.238970 139280796612416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.639537s (batch_requests_ready=0.255522s, total_serialization_initiated=0.383946s, others=0.000069s) I0423 17:43:17.239090 139280796612416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.643552s (all_items=0.000020s, per_item={'items': '0.00002003'}, temp_paths=0.643532) I0423 17:43:17.240081 139158307710720 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:43:17.240224 139280796612416 async_checkpointer.py:561] Finished blocking save. Time taken: 0.980029s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/0. I0423 17:43:17.668370 139280796612416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:43:17.668724 139154046314240 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:43:17.668890 139280796612416 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776966196.2578688, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776966196.2601645, 'checkpointer_blocking_duration_secs': 0.9802463054656982, 'get_old_steps_start_time': 1776966197.24043, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776966196.2561133, 'checkpoint_manager_blocking_duration_secs': 1.4127345085144043} I0423 17:43:17.669012 139280796612416 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0423 17:43:17.669108 139280796612416 max_utils.py:750] Memstats: After params initialized: I0423 17:43:17.669167 139280796612416 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_24(process=6,(0,6,0,0)) I0423 17:43:17.669200 139280796612416 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_25(process=6,(1,6,0,0)) I0423 17:43:17.669234 139280796612416 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_28(process=6,(0,7,0,0)) I0423 17:43:17.669260 139280796612416 max_utils.py:756] Using (GB) 0.45 / 31.25 (1.440000%) on TPU_29(process=6,(1,7,0,0)) I0423 17:43:17.934529 139280796612416 metric_logger.py:185] completed step: 0, seconds: 13.335, TFLOP/s/device: 1.019, Tokens/s/device: 153.583, total_weights: 65536, loss: 10.872 I0423 17:43:18.116720 139280796612416 metric_logger.py:185] completed step: 1, seconds: 1.837, TFLOP/s/device: 7.395, Tokens/s/device: 1114.703, total_weights: 65536, loss: 10.872 I0423 17:43:18.569388 139280796612416 metric_logger.py:185] completed step: 2, seconds: 0.031, TFLOP/s/device: 431.982, Tokens/s/device: 65113.026, total_weights: 65536, loss: 10.856 I0423 17:43:18.724658 139280796612416 metric_logger.py:185] completed step: 3, seconds: 0.430, TFLOP/s/device: 31.608, Tokens/s/device: 4764.364, total_weights: 65536, loss: 10.824 I0423 17:43:19.036144 139280796612416 metric_logger.py:185] completed step: 4, seconds: 0.184, TFLOP/s/device: 73.781, Tokens/s/device: 11121.127, total_weights: 65536, loss: 10.793 I0423 17:43:19.045689 139280796612416 metric_logger.py:185] completed step: 5, seconds: 0.155, TFLOP/s/device: 87.667, Tokens/s/device: 13214.097, total_weights: 65536, loss: 10.763 I0423 17:43:39.251628 2608 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 17:43:39.434605 139280796612416 metric_logger.py:185] completed step: 6, seconds: 0.312, TFLOP/s/device: 43.529, Tokens/s/device: 6561.116, total_weights: 65536, loss: 10.733 I0423 17:43:39.589804 139280796612416 metric_logger.py:185] completed step: 7, seconds: 20.227, TFLOP/s/device: 0.672, Tokens/s/device: 101.252, total_weights: 65536, loss: 10.706 I0423 17:43:39.744950 139280796612416 metric_logger.py:185] completed step: 8, seconds: 0.171, TFLOP/s/device: 79.348, Tokens/s/device: 11960.172, total_weights: 65536, loss: 10.680 I0423 17:43:39.752984 139280796612416 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0423 17:43:40.055791 139280796612416 checkpointing.py:798] Waited 0.3027794361114502 seconds for step 10 to finish before starting checkpointing. I0423 17:43:40.058201 139280796612416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:43:41.169421 139158316103424 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/0/items/array_metadatas/process_6 I0423 17:43:56.090798 139158307710720 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 40.003 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 39.49131536483765 s) (per-host) I0423 17:43:56.090915 139158307710720 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.850607s. I0423 17:44:04.260247 139158307710720 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 47.019923s. I0423 17:44:04.260541 139154046314240 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:44:04.260665 139154046314240 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:44:04.260718 139154046314240 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:44:04.263453 139154046314240 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 17:44:04.263632 139280796612416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0423 17:44:04.263761 139280796612416 checkpoint_manager.py:1458] Waiting for previous save to complete took 24.205564 seconds. If this number is high, consider checkpointing less frequently. I0423 17:44:04.265541 139280796612416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 10 I0423 17:44:04.268984 139280796612416 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/10. I0423 17:44:04.815053 139280796612416 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0423 17:44:04.815179 139280796612416 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 I0423 17:44:04.869801 139280796612416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.058089s I0423 17:44:04.869969 139280796612416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.766 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3237128257751465 s) (per-host) I0423 17:44:04.870021 139280796612416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.323774s (batch_requests_ready=0.256103s, total_serialization_initiated=0.067602s, others=0.000069s) I0423 17:44:04.870147 139280796612416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.327806s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.327790) I0423 17:44:04.871083 139151895000832 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:44:04.871260 139280796612416 async_checkpointer.py:561] Finished blocking save. Time taken: 0.605649s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/10. I0423 17:44:04.886990 139280796612416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:44:04.887263 139154046314240 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:44:04.887397 139280796612416 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776966220.058172, 'wait_for_prev_duration_secs': 24.20556378364563, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776966244.2655797, 'checkpointer_blocking_duration_secs': 0.6057925224304199, 'get_old_steps_start_time': 1776966244.8713892, 'get_old_steps_duration_secs': 2.4557113647460938e-05, 'checkpoint_manager_blocking_start_time': 1776966220.0562453, 'checkpoint_manager_blocking_duration_secs': 24.831114053726196} I0423 17:44:04.887525 139280796612416 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0423 17:44:04.887926 139280796612416 metric_logger.py:185] completed step: 9, seconds: 0.155, TFLOP/s/device: 87.556, Tokens/s/device: 13197.322, total_weights: 65536, loss: 10.656 I0423 17:44:04.898922 139280796612416 metric_logger.py:185] completed step: 10, seconds: 0.155, TFLOP/s/device: 87.659, Tokens/s/device: 13212.903, total_weights: 65536, loss: 10.636 I0423 17:44:05.049913 139280796612416 metric_logger.py:185] completed step: 11, seconds: 25.143, TFLOP/s/device: 0.540, Tokens/s/device: 81.453, total_weights: 65536, loss: 10.618 I0423 17:44:05.205156 139280796612416 metric_logger.py:185] completed step: 12, seconds: 0.010, TFLOP/s/device: 1387.008, Tokens/s/device: 209064.924, total_weights: 65536, loss: 10.602 I0423 17:44:05.360428 139280796612416 metric_logger.py:185] completed step: 13, seconds: 0.152, TFLOP/s/device: 89.267, Tokens/s/device: 13455.360, total_weights: 65536, loss: 10.588 I0423 17:44:05.515442 139280796612416 metric_logger.py:185] completed step: 14, seconds: 0.155, TFLOP/s/device: 87.442, Tokens/s/device: 13180.250, total_weights: 65536, loss: 10.577 I0423 17:44:05.670517 139280796612416 metric_logger.py:185] completed step: 15, seconds: 0.155, TFLOP/s/device: 87.556, Tokens/s/device: 13197.407, total_weights: 65536, loss: 10.568 I0423 17:44:05.825570 139280796612416 metric_logger.py:185] completed step: 16, seconds: 0.155, TFLOP/s/device: 87.639, Tokens/s/device: 13209.920, total_weights: 65536, loss: 10.562 I0423 17:44:05.980763 139280796612416 metric_logger.py:185] completed step: 17, seconds: 0.155, TFLOP/s/device: 87.708, Tokens/s/device: 13220.324, total_weights: 65536, loss: 10.556 I0423 17:44:06.135776 139280796612416 metric_logger.py:185] completed step: 18, seconds: 0.155, TFLOP/s/device: 87.575, Tokens/s/device: 13200.299, total_weights: 65536, loss: 10.552 I0423 17:44:06.293228 139280796612416 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0423 17:44:06.294368 139280796612416 checkpointing.py:798] Waited 0.0011518001556396484 seconds for step 19 to finish before starting checkpointing. I0423 17:44:06.296553 139280796612416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:44:10.172200 139156697097984 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/10/items/array_metadatas/process_6 I0423 17:44:25.767466 139151895000832 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 74.443 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.221164226531982 s) (per-host) I0423 17:44:25.767589 139151895000832 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.896379s. I0423 17:44:33.318281 139151895000832 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 28.447054s. I0423 17:44:33.318572 139154046314240 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:44:33.318691 139154046314240 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:44:33.318741 139154046314240 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:44:33.320382 139154046314240 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0423 17:44:33.320561 139280796612416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0423 17:44:33.320680 139280796612416 checkpoint_manager.py:1458] Waiting for previous save to complete took 27.024139 seconds. If this number is high, consider checkpointing less frequently. I0423 17:44:33.322248 139280796612416 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 19 I0423 17:44:33.325658 139280796612416 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/19. I0423 17:44:34.378670 139280796612416 jax_array_handlers.py:358] Scheduling D2H of 153 prioritized jax.Array. I0423 17:44:34.378779 139280796612416 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 I0423 17:44:34.435991 139280796612416 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.060730s I0423 17:44:34.436167 139280796612416 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.011 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7672498226165771 s) (per-host) I0423 17:44:34.436216 139280796612416 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.767309s (batch_requests_ready=0.697053s, total_serialization_initiated=0.070192s, others=0.000064s) I0423 17:44:34.436335 139280796612416 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.771264s (all_items=0.000010s, per_item={'items': '0.00000978'}, temp_paths=0.771254) I0423 17:44:34.437201 139154574792448 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0423 17:44:34.437364 139280796612416 async_checkpointer.py:561] Finished blocking save. Time taken: 1.115034s. Continuing background save to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/19. I0423 17:44:34.442764 139280796612416 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0423 17:44:34.443052 139154046314240 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 17:44:34.443202 139280796612416 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776966246.2965257, 'wait_for_prev_duration_secs': 27.024138927459717, 'time_between_consecutive_saves_sec': 2.033034324645996, 'checkpointer_blocking_start_time': 1776966273.3222985, 'checkpointer_blocking_duration_secs': 1.1151740550994873, 'get_old_steps_start_time': 1776966274.437487, 'get_old_steps_duration_secs': 2.0742416381835938e-05, 'checkpoint_manager_blocking_start_time': 1776966246.2946787, 'checkpoint_manager_blocking_duration_secs': 28.148491859436035} I0423 17:44:34.443384 139280796612416 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0423 17:44:34.443428 139280796612416 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 17:44:39.742632 139154583185152 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 153 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_set_defaults_true_20260423_155242/nnx_xpk_feat_nnx_set_defaults_true_20260423_155242_05_fp8/checkpoints/19/items/array_metadatas/process_6 I0423 17:44:55.149191 139154574792448 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 73.545 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.480229377746582 s) (per-host) I0423 17:44:55.149313 139154574792448 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.711997s. I0423 17:45:00.966085 139154574792448 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 26.528754s. I0423 17:45:00.966416 139154046314240 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 17:45:00.966534 139154046314240 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 17:45:00.966575 139154046314240 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0423 17:45:00.967938 139154046314240 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0423 17:45:00.968165 139280796612416 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0423 17:45:00.968348 139280796612416 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 17:45:00.968970 139280796612416 metric_logger.py:185] completed step: 19, seconds: 0.155, TFLOP/s/device: 87.682, Tokens/s/device: 13216.314, 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: Thu Apr 23 17:45:13 UTC 2026 EXIT_CODE=0