MaxView

XPK Start: Wed Apr 22 11:17:22 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-22 11:17:46.116869: 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)
I0422 11:17:46.656951 135522405529408 max_utils.py:273] Attempting to initialize the jax distributed system...
I0422 11:17:55.700233 135522405529408 distributed.py:149] Starting JAX distributed service on [::]:8482
I0422 11:17:55.702642 135522405529408 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-260rp-slice-job-0-0.mt-05-fp8-260rp:8482
I0422 11:17:57.512109 135522405529408 max_utils.py:284] Jax distributed system initialized!
I0422 11:18:03.970217 135522405529408 max_utils.py:800] System Information: Jax Version: 0.9.2
I0422 11:18:03.970324 135522405529408 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0422 11:18:03.970366 135522405529408 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0422 11:18:03.970408 135522405529408 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0422 11:18:04.664135 135522405529408 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0422 11:18:05.101792 135522405529408 checkpointing.py:688] Setting up checkpoint logger...
I0422 11:18:05.101915 135522405529408 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0422 11:18:05.101969 135522405529408 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0422 11:18:05.102239 135522405529408 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7b410d44b7a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0422 11:18:07.900349 135522405529408 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0422 11:18:07.900605 135522405529408 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b40012672f0>}, handler_registry=None
I0422 11:18:07.900887 135522405529408 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b40012672f0>` 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`.
I0422 11:18:07.900945 135522405529408 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b2bf45e4710>` 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`.
I0422 11:18:07.900993 135522405529408 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 0x7b40012672f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b40012672f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b2bf45e4710>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b2bf45e4710>}).
I0422 11:18:07.901375 135522405529408 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0422 11:18:07.901459 135522405529408 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b2c181a7b00> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0422 11:18:09.063348 135522405529408 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints
I0422 11:18:09.421838 135522405529408 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_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b2c181dae10>
I0422 11:18:09.422018 135522405529408 checkpointing.py:302] Checkpoint manager created!
I0422 11:18:10.712134 135522405529408 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0422 11:18:10.712256 135522405529408 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32

I0422 11:18:13.239011 135522405529408 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 11:18:13.239117 135522405529408 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 11:18:13.244655 135522405529408 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 11:18:13.244714 135522405529408 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 11:18:13.261248 135522405529408 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 11:18:13.261313 135522405529408 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 11:18:13.277053 135522405529408 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 11:18:13.277121 135522405529408 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 11:18:13.341623 135522405529408 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 11:18:13.341718 135522405529408 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 11:18:13.357754 135522405529408 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 11:18:13.357812 135522405529408 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 11:18:13.373770 135522405529408 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 11:18:13.373829 135522405529408 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 11:18:13.404008 135522405529408 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0422 11:18:13.404081 135522405529408 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 11:18:13.464077 135522405529408 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0422 11:18:13.464171 135522405529408 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0422 11:18:27.937748 135522405529408 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.
I0422 11:18:27.941599 135522405529408 metric_logger.py:301] number parameters: 1.104 billion
I0422 11:18:41.731675 135522405529408 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0422 11:18:41.894175 135522405529408 checkpointing.py:798] Waited 0.16248559951782227 seconds for step 0 to finish before starting checkpointing.
I0422 11:18:41.898689 135522405529408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 11:18:41.900052 135522405529408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0422 11:18:41.901550 135522405529408 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0.
I0422 11:18:42.245118 135522405529408 signaling_client.py:364] Using JaxDistributedSignalingClient
I0422 11:18:42.246049 135522405529408 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0422 11:18:42.246134 135522405529408 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
I0422 11:18:42.656935 135522405529408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.413508s
I0422 11:18:42.657134 135522405529408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 3.578 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.43114542961120605 s) (per-host)
I0422 11:18:42.657186 135522405529408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.431234s (batch_requests_ready=0.005609s, total_serialization_initiated=0.425529s, others=0.000097s)
I0422 11:18:42.657312 135522405529408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.435251s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.435234)
I0422 11:18:42.658126 135522405529408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0.
I0422 11:18:42.658426 135393475294976 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 11:38:42.658394
I0422 11:18:42.666173 135522405529408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 11:18:42.666414 135392938424064 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 11:18:42.666537 135522405529408 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776856721.898668, 'wait_for_prev_duration_secs': 6.4849853515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776856721.900089, 'checkpointer_blocking_duration_secs': 0.7584993839263916, 'get_old_steps_start_time': 1776856722.6586132, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776856721.894673, 'checkpoint_manager_blocking_duration_secs': 0.7718255519866943}
I0422 11:18:42.666647 135522405529408 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0422 11:18:42.666724 135522405529408 max_utils.py:750] 
Memstats: After params initialized:
I0422 11:18:42.666771 135522405529408 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_24(process=6,(0,6,0,0))
I0422 11:18:42.666803 135522405529408 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_25(process=6,(1,6,0,0))
I0422 11:18:42.666829 135522405529408 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_28(process=6,(0,7,0,0))
I0422 11:18:42.666853 135522405529408 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_29(process=6,(1,7,0,0))
I0422 11:18:42.984755 135522405529408 metric_logger.py:196] completed step: 0, seconds: 13.788, TFLOP/s/device: 0.985, Tokens/s/device: 148.536, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0422 11:18:43.166349 135522405529408 metric_logger.py:196] completed step: 1, seconds: 1.250, TFLOP/s/device: 10.867, Tokens/s/device: 1637.972, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0422 11:18:43.643873 135522405529408 metric_logger.py:196] completed step: 2, seconds: 0.030, TFLOP/s/device: 458.019, Tokens/s/device: 69037.586, total_weights: 65536, loss: 10.856, lm_loss: 10.856, perplexity: 51864.762
I0422 11:18:43.801322 135522405529408 metric_logger.py:196] completed step: 3, seconds: 0.454, TFLOP/s/device: 29.951, Tokens/s/device: 4514.523, total_weights: 65536, loss: 10.824, lm_loss: 10.824, perplexity: 50225.512
I0422 11:18:44.116145 135522405529408 metric_logger.py:196] completed step: 4, seconds: 0.186, TFLOP/s/device: 72.856, Tokens/s/device: 10981.645, total_weights: 65536, loss: 10.793, lm_loss: 10.793, perplexity: 48689.156
I0422 11:18:44.125585 135522405529408 metric_logger.py:196] completed step: 5, seconds: 0.157, TFLOP/s/device: 86.288, Tokens/s/device: 13006.230, total_weights: 65536, loss: 10.763, lm_loss: 10.763, perplexity: 47229.012
I0422 11:18:45.435880    2838 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0422 11:18:47.458534 135393421981440 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_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/0/items/array_metadatas/process_6
I0422 11:19:02.234473 135393475294976 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 78.955 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 20.00847339630127 s) (per-host)
I0422 11:19:02.234597 135393475294976 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 19.576057s.
I0422 11:19:07.890289 135522405529408 metric_logger.py:196] completed step: 6, seconds: 0.315, TFLOP/s/device: 43.165, Tokens/s/device: 6506.317, total_weights: 65536, loss: 10.733, lm_loss: 10.733, perplexity: 45864.258
I0422 11:19:08.047122 135522405529408 metric_logger.py:196] completed step: 7, seconds: 23.614, TFLOP/s/device: 0.575, Tokens/s/device: 86.730, total_weights: 65536, loss: 10.706, lm_loss: 10.706, perplexity: 44620.242
I0422 11:19:08.204331 135522405529408 metric_logger.py:196] completed step: 8, seconds: 0.160, TFLOP/s/device: 84.718, Tokens/s/device: 12769.592, total_weights: 65536, loss: 10.680, lm_loss: 10.680, perplexity: 43477.055
I0422 11:19:08.212443 135522405529408 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0422 11:19:08.518373 135522405529408 checkpointing.py:798] Waited 0.3059051036834717 seconds for step 10 to finish before starting checkpointing.
I0422 11:19:08.520787 135522405529408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 11:19:10.883632 135393475294976 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 28.225076s.
I0422 11:19:10.883840 135392938424064 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 11:19:10.883908 135392938424064 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 11:19:10.883958 135392938424064 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0422 11:19:10.887004 135392938424064 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0422 11:19:10.887188 135522405529408 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0422 11:19:10.887331 135522405529408 checkpoint_manager.py:1452] Waiting for previous save to complete took 2.366547 seconds. If this number is high, consider checkpointing less frequently.
I0422 11:19:10.889323 135522405529408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0422 11:19:10.891225 135522405529408 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10.
I0422 11:19:11.618958 135522405529408 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0422 11:19:11.619063 135522405529408 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
I0422 11:19:11.674882 135522405529408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.058401s
I0422 11:19:11.675042 135522405529408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 21.097 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.07312655448913574 s) (per-host)
I0422 11:19:11.675091 135522405529408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.073185s (batch_requests_ready=0.005165s, total_serialization_initiated=0.067955s, others=0.000065s)
I0422 11:19:11.675221 135522405529408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.077258s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.077243)
I0422 11:19:11.675942 135522405529408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.79 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10.
I0422 11:19:11.676313 135392938424064 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 11:39:11.676277
I0422 11:19:12.112931 135522405529408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 11:19:12.113315 135391294240512 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 11:19:12.113475 135522405529408 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776856748.520755, 'wait_for_prev_duration_secs': 2.366546630859375, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776856750.8893633, 'checkpointer_blocking_duration_secs': 0.7870595455169678, 'get_old_steps_start_time': 1776856751.6764462, 'get_old_steps_duration_secs': 2.8133392333984375e-05, 'checkpoint_manager_blocking_start_time': 1776856748.5188332, 'checkpoint_manager_blocking_duration_secs': 3.5946035385131836}
I0422 11:19:12.113683 135522405529408 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0422 11:19:12.114449 135522405529408 metric_logger.py:196] completed step: 9, seconds: 0.157, TFLOP/s/device: 86.616, Tokens/s/device: 13055.729, total_weights: 65536, loss: 10.656, lm_loss: 10.656, perplexity: 42467.320
I0422 11:19:12.125762 135522405529408 metric_logger.py:196] completed step: 10, seconds: 0.157, TFLOP/s/device: 86.519, Tokens/s/device: 13041.014, total_weights: 65536, loss: 10.636, lm_loss: 10.636, perplexity: 41601.941
I0422 11:19:12.278700 135522405529408 metric_logger.py:196] completed step: 11, seconds: 3.911, TFLOP/s/device: 3.474, Tokens/s/device: 523.699, total_weights: 65536, loss: 10.618, lm_loss: 10.618, perplexity: 40849.570
I0422 11:19:12.435618 135522405529408 metric_logger.py:196] completed step: 12, seconds: 0.010, TFLOP/s/device: 1411.503, Tokens/s/device: 212757.116, total_weights: 65536, loss: 10.602, lm_loss: 10.602, perplexity: 40203.926
I0422 11:19:13.012310 135522405529408 metric_logger.py:196] completed step: 13, seconds: 0.154, TFLOP/s/device: 88.059, Tokens/s/device: 13273.189, total_weights: 65536, loss: 10.588, lm_loss: 10.588, perplexity: 39652.145
I0422 11:19:13.169000 135522405529408 metric_logger.py:196] completed step: 14, seconds: 0.571, TFLOP/s/device: 23.799, Tokens/s/device: 3587.281, total_weights: 65536, loss: 10.577, lm_loss: 10.577, perplexity: 39212.879
I0422 11:19:13.487719 135522405529408 metric_logger.py:196] completed step: 15, seconds: 0.163, TFLOP/s/device: 83.537, Tokens/s/device: 12591.531, total_weights: 65536, loss: 10.568, lm_loss: 10.568, perplexity: 38879.223
I0422 11:19:13.644241 135522405529408 metric_logger.py:196] completed step: 16, seconds: 0.313, TFLOP/s/device: 43.416, Tokens/s/device: 6544.072, total_weights: 65536, loss: 10.562, lm_loss: 10.562, perplexity: 38619.105
I0422 11:19:13.801150 135522405529408 metric_logger.py:196] completed step: 17, seconds: 0.162, TFLOP/s/device: 83.621, Tokens/s/device: 12604.240, total_weights: 65536, loss: 10.556, lm_loss: 10.556, perplexity: 38395.816
I0422 11:19:13.957983 135522405529408 metric_logger.py:196] completed step: 18, seconds: 0.157, TFLOP/s/device: 86.789, Tokens/s/device: 13081.748, total_weights: 65536, loss: 10.552, lm_loss: 10.552, perplexity: 38236.652
I0422 11:19:14.116587 135522405529408 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0422 11:19:14.117802 135522405529408 checkpointing.py:798] Waited 0.0012235641479492188 seconds for step 19 to finish before starting checkpointing.
I0422 11:19:14.120074 135522405529408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 11:19:17.483593 135391319418624 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_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/10/items/array_metadatas/process_6
I0422 11:19:33.059572 135392938424064 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 73.622 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.457615613937378 s) (per-host)
I0422 11:19:33.059701 135392938424064 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.382717s.
I0422 11:19:39.506698 135392938424064 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 27.829698s.
I0422 11:19:39.506924 135391294240512 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 11:19:39.506982 135391294240512 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 11:19:39.507030 135391294240512 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0422 11:19:39.509017 135391294240512 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0422 11:19:39.509207 135522405529408 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0422 11:19:39.509343 135522405529408 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.389274 seconds. If this number is high, consider checkpointing less frequently.
I0422 11:19:39.510833 135522405529408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0422 11:19:39.512793 135522405529408 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19.
I0422 11:19:39.916511 135522405529408 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0422 11:19:39.916613 135522405529408 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
I0422 11:19:39.974575 135522405529408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.060613s
I0422 11:19:39.974729 135522405529408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 20.656 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.07468771934509277 s) (per-host)
I0422 11:19:39.974778 135522405529408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.074746s (batch_requests_ready=0.005054s, total_serialization_initiated=0.069629s, others=0.000062s)
I0422 11:19:39.974890 135522405529408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.078727s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.078717)
I0422 11:19:39.975507 135522405529408 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.46 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19.
I0422 11:19:39.975843 135390228891392 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 11:39:39.975806
I0422 11:19:39.977812 135522405529408 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 11:19:39.978025 135391294240512 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 11:19:39.978134 135522405529408 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776856754.1200457, 'wait_for_prev_duration_secs': 25.38927388191223, 'time_between_consecutive_saves_sec': 3.2330005168914795, 'checkpointer_blocking_start_time': 1776856779.5108721, 'checkpointer_blocking_duration_secs': 0.4651224613189697, 'get_old_steps_start_time': 1776856779.976012, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776856754.118134, 'checkpoint_manager_blocking_duration_secs': 25.859954833984375}
I0422 11:19:39.978300 135522405529408 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0422 11:19:39.978345 135522405529408 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 11:19:45.052163 135389717198592 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_trainstate_and_training_loop_20260422_093106/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260422_093106_05_fp8/checkpoints/19/items/array_metadatas/process_6
I0422 11:20:01.287485 135390228891392 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 73.864 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.387400150299072 s) (per-host)
I0422 11:20:01.287614 135390228891392 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.311655s.
I0422 11:20:07.919907 135390228891392 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 27.943932s.
I0422 11:20:07.920232 135391294240512 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 11:20:07.920362 135391294240512 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 11:20:07.920403 135391294240512 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0422 11:20:07.922159 135391294240512 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0422 11:20:07.922339 135522405529408 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0422 11:20:07.922518 135522405529408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 11:20:07.923449 135522405529408 metric_logger.py:196] completed step: 19, seconds: 0.157, TFLOP/s/device: 86.640, Tokens/s/device: 13059.392, total_weights: 65536, loss: 10.549, lm_loss: 10.549, perplexity: 38125.074
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Wed Apr 22 11:20:17 UTC 2026
EXIT_CODE=0