MaxView

XPK Start: Fri Apr 24 10:50:34 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
`rope_parameters`'s factor field must be a float >= 1, got 40
`rope_parameters`'s beta_fast field must be a float, got 32
`rope_parameters`'s beta_slow field must be a float, got 1
DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 
2026-04-24 10:50:59.598540: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)
I0424 10:50:59.810040 138356317857600 max_utils.py:273] Attempting to initialize the jax distributed system...
I0424 10:51:08.851871 138356317857600 distributed.py:149] Starting JAX distributed service on [::]:8482
I0424 10:51:08.854360 138356317857600 distributed.py:172] Connecting to JAX distributed service on mt-05-fp8-8qznm-slice-job-0-0.mt-05-fp8-8qznm:8482
I0424 10:51:10.024068 138356317857600 max_utils.py:284] Jax distributed system initialized!
I0424 10:51:16.088742 138356317857600 max_utils.py:800] System Information: Jax Version: 0.9.2
I0424 10:51:16.088847 138356317857600 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0424 10:51:16.088890 138356317857600 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
I0424 10:51:16.088927 138356317857600 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0424 10:51:16.783276 138356317857600 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0424 10:51:17.213359 138356317857600 checkpointing.py:688] Setting up checkpoint logger...
I0424 10:51:17.213484 138356317857600 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0424 10:51:17.213533 138356317857600 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0424 10:51:17.213748 138356317857600 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 0x7dd4df644080>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0424 10:51:20.152343 138356317857600 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0424 10:51:20.152579 138356317857600 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 0x7dd3d36e61b0>}, handler_registry=None
I0424 10:51:20.152821 138356317857600 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dd3d36e61b0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`.
I0424 10:51:20.152870 138356317857600 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc0b40c9ca0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`.
I0424 10:51:20.152905 138356317857600 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 0x7dd3d36e61b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dd3d36e61b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc0b40c9ca0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dc0b40c9ca0>}).
I0424 10:51:20.153252 138356317857600 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0424 10:51:20.153329 138356317857600 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7dc09c396480> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0424 10:51:20.847479 138356317857600 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints
I0424 10:51:20.953644 138356317857600 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_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dc09c3418b0>
I0424 10:51:20.953822 138356317857600 checkpointing.py:302] Checkpoint manager created!
I0424 10:51:22.226399 138356317857600 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0424 10:51:22.226524 138356317857600 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32

I0424 10:51:24.747212 138356317857600 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0424 10:51:24.747318 138356317857600 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 10:51:24.752780 138356317857600 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0424 10:51:24.752839 138356317857600 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 10:51:24.769594 138356317857600 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn').
I0424 10:51:24.769653 138356317857600 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 10:51:24.785628 138356317857600 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn').
I0424 10:51:24.785687 138356317857600 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 10:51:24.845502 138356317857600 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 10:51:24.845587 138356317857600 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 10:51:24.861864 138356317857600 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 10:51:24.861924 138356317857600 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 10:51:24.878050 138356317857600 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 10:51:24.878122 138356317857600 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 10:51:24.909505 138356317857600 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv').
I0424 10:51:24.909581 138356317857600 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 10:51:24.969449 138356317857600 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0424 10:51:24.969530 138356317857600 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0424 10:51:38.950857 138356317857600 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0424 10:51:38.954704 138356317857600 metric_logger.py:301] number parameters: 1.104 billion
I0424 10:51:52.629630 138356317857600 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0424 10:51:52.791331 138356317857600 checkpointing.py:798] Waited 0.16168975830078125 seconds for step 0 to finish before starting checkpointing.
I0424 10:51:52.793605 138356317857600 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0424 10:51:52.795631 138356317857600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0424 10:51:52.797002 138356317857600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/0.
I0424 10:51:53.156077 138356317857600 signaling_client.py:364] Using JaxDistributedSignalingClient
I0424 10:51:53.157035 138356317857600 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0424 10:51:53.157119 138356317857600 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0424 10:51:53.566447 138356317857600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.412023s
I0424 10:51:53.566618 138356317857600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 3.596 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.42904067039489746 s) (per-host)
I0424 10:51:53.566669 138356317857600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.429102s (batch_requests_ready=0.005949s, total_serialization_initiated=0.423084s, others=0.000070s)
I0424 10:51:53.566786 138356317857600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.433228s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.433210)
I0424 10:51:53.567530 138356317857600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.77 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/0.
I0424 10:51:53.567888 138227429209856 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 11:11:53.567849
I0424 10:51:53.587714 138356317857600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0424 10:51:53.588035 138226928965376 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0424 10:51:53.588217 138356317857600 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777027912.7935858, 'wait_for_prev_duration_secs': 6.389617919921875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777027912.7956705, 'checkpointer_blocking_duration_secs': 0.7723715305328369, 'get_old_steps_start_time': 1777027913.5680685, 'get_old_steps_duration_secs': 4.4345855712890625e-05, 'checkpoint_manager_blocking_start_time': 1777027912.791789, 'checkpoint_manager_blocking_duration_secs': 0.7963881492614746}
I0424 10:51:53.588331 138356317857600 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0424 10:51:53.588410 138356317857600 max_utils.py:750] 
Memstats: After params initialized:
I0424 10:51:53.588459 138356317857600 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_24(process=6,(0,6,0,0))
I0424 10:51:53.588490 138356317857600 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_25(process=6,(1,6,0,0))
I0424 10:51:53.588516 138356317857600 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_28(process=6,(0,7,0,0))
I0424 10:51:53.588542 138356317857600 max_utils.py:756] 	Using (GB) 0.45 / 31.25 (1.440000%) on TPU_29(process=6,(1,7,0,0))
I0424 10:51:53.910381 138356317857600 metric_logger.py:196] completed step: 0, seconds: 13.673, TFLOP/s/device: 0.994, Tokens/s/device: 149.787, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0424 10:51:54.090132 138356317857600 metric_logger.py:196] completed step: 1, seconds: 1.278, TFLOP/s/device: 10.632, Tokens/s/device: 1602.529, total_weights: 65536, loss: 10.872, lm_loss: 10.872, perplexity: 52680.742
I0424 10:51:54.552425 138356317857600 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 509.549, Tokens/s/device: 76804.800, total_weights: 65536, loss: 10.856, lm_loss: 10.856, perplexity: 51864.762
I0424 10:51:54.709357 138356317857600 metric_logger.py:196] completed step: 3, seconds: 0.439, TFLOP/s/device: 30.928, Tokens/s/device: 4661.761, total_weights: 65536, loss: 10.824, lm_loss: 10.824, perplexity: 50225.512
I0424 10:51:55.024064 138356317857600 metric_logger.py:196] completed step: 4, seconds: 0.187, TFLOP/s/device: 72.832, Tokens/s/device: 10977.996, total_weights: 65536, loss: 10.793, lm_loss: 10.793, perplexity: 48689.156
I0424 10:51:55.033643 138356317857600 metric_logger.py:196] completed step: 5, seconds: 0.158, TFLOP/s/device: 86.077, Tokens/s/device: 12974.425, total_weights: 65536, loss: 10.763, lm_loss: 10.763, perplexity: 47229.012
I0424 10:51:56.409976    2807 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0424 10:51:58.423775 138226937358080 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_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/0/items/array_metadatas/process_6
I0424 10:52:11.302725 138227429209856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 86.967 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 18.16510772705078 s) (per-host)
I0424 10:52:11.302850 138227429209856 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 17.734847s.
I0424 10:52:19.301129 138356317857600 metric_logger.py:196] completed step: 6, seconds: 0.314, TFLOP/s/device: 43.254, Tokens/s/device: 6519.656, total_weights: 65536, loss: 10.733, lm_loss: 10.733, perplexity: 45864.258
I0424 10:52:19.310124 138356317857600 metric_logger.py:196] completed step: 7, seconds: 24.267, TFLOP/s/device: 0.560, Tokens/s/device: 84.395, total_weights: 65536, loss: 10.706, lm_loss: 10.706, perplexity: 44620.242
I0424 10:52:19.471200 138356317857600 metric_logger.py:196] completed step: 8, seconds: 0.010, TFLOP/s/device: 1375.912, Tokens/s/device: 207392.405, total_weights: 65536, loss: 10.680, lm_loss: 10.680, perplexity: 43477.055
I0424 10:52:19.479756 138356317857600 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0424 10:52:19.785209 138356317857600 checkpointing.py:798] Waited 0.30542707443237305 seconds for step 10 to finish before starting checkpointing.
I0424 10:52:19.787566 138356317857600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0424 10:52:22.468291 138227429209856 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 28.900270s.
I0424 10:52:22.468622 138226928965376 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0424 10:52:22.468751 138226928965376 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0424 10:52:22.468803 138226928965376 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0424 10:52:22.471380 138226928965376 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0424 10:52:22.471608 138356317857600 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.
W0424 10:52:22.471741 138356317857600 checkpoint_manager.py:1452] Waiting for previous save to complete took 2.684175 seconds. If this number is high, consider checkpointing less frequently.
I0424 10:52:22.473688 138356317857600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0424 10:52:22.475758 138356317857600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/10.
I0424 10:52:22.770488 138356317857600 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0424 10:52:22.770602 138356317857600 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0424 10:52:22.819383 138356317857600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.051512s
I0424 10:52:22.819534 138356317857600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 23.171 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06658172607421875 s) (per-host)
I0424 10:52:22.819595 138356317857600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.066643s (batch_requests_ready=0.005110s, total_serialization_initiated=0.061465s, others=0.000068s)
I0424 10:52:22.819718 138356317857600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.070545s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.070529)
I0424 10:52:22.820460 138356317857600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/10.
I0424 10:52:22.820787 138226928965376 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 11:12:22.820753
I0424 10:52:22.822828 138356317857600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0424 10:52:22.823140 138223707739904 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0424 10:52:22.823283 138356317857600 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777027939.7875364, 'wait_for_prev_duration_secs': 2.6841747760772705, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777027942.4737275, 'checkpointer_blocking_duration_secs': 0.3471996784210205, 'get_old_steps_start_time': 1777027942.8209515, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1777027939.7856557, 'checkpoint_manager_blocking_duration_secs': 3.037593126296997}
I0424 10:52:22.823457 138356317857600 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0424 10:52:22.824066 138356317857600 metric_logger.py:196] completed step: 9, seconds: 0.008, TFLOP/s/device: 1683.659, Tokens/s/device: 253779.430, total_weights: 65536, loss: 10.656, lm_loss: 10.656, perplexity: 42467.320
I0424 10:52:22.835262 138356317857600 metric_logger.py:196] completed step: 10, seconds: 0.163, TFLOP/s/device: 83.612, Tokens/s/device: 12602.922, total_weights: 65536, loss: 10.636, lm_loss: 10.636, perplexity: 41601.941
I0424 10:52:22.988421 138356317857600 metric_logger.py:196] completed step: 11, seconds: 3.353, TFLOP/s/device: 4.053, Tokens/s/device: 610.881, total_weights: 65536, loss: 10.618, lm_loss: 10.618, perplexity: 40849.570
I0424 10:52:23.576085 138356317857600 metric_logger.py:196] completed step: 12, seconds: 0.010, TFLOP/s/device: 1338.633, Tokens/s/device: 201773.399, total_weights: 65536, loss: 10.602, lm_loss: 10.602, perplexity: 40203.926
I0424 10:52:23.732794 138356317857600 metric_logger.py:196] completed step: 13, seconds: 0.579, TFLOP/s/device: 23.470, Tokens/s/device: 3537.677, total_weights: 65536, loss: 10.588, lm_loss: 10.588, perplexity: 39652.145
I0424 10:52:23.889683 138356317857600 metric_logger.py:196] completed step: 14, seconds: 0.163, TFLOP/s/device: 83.372, Tokens/s/device: 12566.730, total_weights: 65536, loss: 10.577, lm_loss: 10.577, perplexity: 39212.879
I0424 10:52:24.046464 138356317857600 metric_logger.py:196] completed step: 15, seconds: 0.157, TFLOP/s/device: 86.701, Tokens/s/device: 13068.475, total_weights: 65536, loss: 10.568, lm_loss: 10.568, perplexity: 38879.223
I0424 10:52:24.203089 138356317857600 metric_logger.py:196] completed step: 16, seconds: 0.157, TFLOP/s/device: 86.633, Tokens/s/device: 13058.226, total_weights: 65536, loss: 10.562, lm_loss: 10.562, perplexity: 38619.105
I0424 10:52:24.359872 138356317857600 metric_logger.py:196] completed step: 17, seconds: 0.157, TFLOP/s/device: 86.460, Tokens/s/device: 13032.218, total_weights: 65536, loss: 10.556, lm_loss: 10.556, perplexity: 38395.816
I0424 10:52:24.519655 138356317857600 metric_logger.py:196] completed step: 18, seconds: 0.156, TFLOP/s/device: 86.943, Tokens/s/device: 13104.935, total_weights: 65536, loss: 10.552, lm_loss: 10.552, perplexity: 38236.652
I0424 10:52:24.678529 138356317857600 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0424 10:52:24.679819 138356317857600 checkpointing.py:798] Waited 0.0012989044189453125 seconds for step 19 to finish before starting checkpointing.
I0424 10:52:24.681832 138356317857600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0424 10:52:28.980892 138223162476288 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_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/10/items/array_metadatas/process_6
I0424 10:52:43.829944 138226928965376 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 74.952 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 21.076953172683716 s) (per-host)
I0424 10:52:43.830072 138226928965376 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 21.009176s.
I0424 10:52:51.884918 138226928965376 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 29.064006s.
I0424 10:52:51.885196 138223707739904 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0424 10:52:51.885313 138223707739904 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0424 10:52:51.885361 138223707739904 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0424 10:52:51.887916 138223707739904 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0424 10:52:51.888116 138356317857600 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.
W0424 10:52:51.888240 138356317857600 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.206411 seconds. If this number is high, consider checkpointing less frequently.
I0424 10:52:51.889804 138356317857600 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0424 10:52:51.891927 138356317857600 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/19.
I0424 10:52:52.592979 138356317857600 jax_array_handlers.py:360] Scheduling D2H of 153 prioritized jax.Array.
I0424 10:52:52.593146 138356317857600 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0424 10:52:52.643645 138356317857600 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.053203s
I0424 10:52:52.643784 138356317857600 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 22.892 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.06739306449890137 s) (per-host)
I0424 10:52:52.643831 138356317857600 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.067451s (batch_requests_ready=0.005112s, total_serialization_initiated=0.062277s, others=0.000062s)
I0424 10:52:52.643908 138356317857600 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.071494s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.071482)
I0424 10:52:52.644544 138356317857600 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/19.
I0424 10:52:52.644865 138223179261696 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 11:12:52.644828
I0424 10:52:53.086300 138356317857600 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0424 10:52:53.086682 138232280889088 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0424 10:52:53.086854 138356317857600 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777027944.6818037, 'wait_for_prev_duration_secs': 27.206411361694336, 'time_between_consecutive_saves_sec': 2.2103240489959717, 'checkpointer_blocking_start_time': 1777027971.8898432, 'checkpointer_blocking_duration_secs': 0.7551741600036621, 'get_old_steps_start_time': 1777027972.6450338, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1777027944.680156, 'checkpoint_manager_blocking_duration_secs': 28.406663179397583}
I0424 10:52:53.086974 138356317857600 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0424 10:52:53.087018 138356317857600 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0424 10:52:58.002168 138224218617600 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_20260424_091312/nnx_xpk_feat_nnx_trainstate_and_training_loop_20260424_091312_05_fp8/checkpoints/19/items/array_metadatas/process_6
I0424 10:53:13.151994 138223179261696 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 76.779 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 20.57555627822876 s) (per-host)
I0424 10:53:13.152118 138223179261696 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 20.507137s.
I0424 10:53:20.797919 138223179261696 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 28.152923s.
I0424 10:53:20.798166 138232280889088 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0424 10:53:20.798220 138232280889088 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0424 10:53:20.798269 138232280889088 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0424 10:53:20.799982 138232280889088 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0424 10:53:20.800163 138356317857600 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.
I0424 10:53:20.800355 138356317857600 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0424 10:53:20.801279 138356317857600 metric_logger.py:196] completed step: 19, seconds: 0.157, TFLOP/s/device: 86.581, Tokens/s/device: 13050.488, 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: Fri Apr 24 10:53:29 UTC 2026
EXIT_CODE=0