XPK Start: Wed Apr 22 11:19:00 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. 2026-04-22 11:19:25.055991: 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 11:19:25.266512 137474987497280 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 11:19:34.306572 137474987497280 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 11:19:34.309083 137474987497280 distributed.py:172] Connecting to JAX distributed service on mt-06-grad-accum-siyfw-slice-job-0-0.mt-06-grad-accum-siyfw:8482 I0422 11:19:35.199842 137474987497280 max_utils.py:284] Jax distributed system initialized! I0422 11:19:40.030188 137474987497280 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 11:19:40.030295 137474987497280 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 11:19:40.030337 137474987497280 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 11:19:40.030373 137474987497280 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 11:19:40.716058 137474987497280 maxtext_utils.py:1732] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 11:19:40.829377 137474987497280 checkpointing.py:688] Setting up checkpoint logger... I0422 11:19:40.829497 137474987497280 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 11:19:40.829548 137474987497280 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 11:19:40.829742 137474987497280 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 0x7d07af3757c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 11:19:43.695739 137474987497280 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 11:19:43.696038 137474987497280 checkpoint_manager.py:708] [process=4][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cf3f8264740>}, handler_registry=None I0422 11:19:43.696280 137474987497280 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cf3f8264740>` 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 11:19:43.696328 137474987497280 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cf3f82663f0>` 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 11:19:43.696364 137474987497280 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 0x7cf3f8264740>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cf3f8264740>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cf3f82663f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cf3f82663f0>}). I0422 11:19:43.696697 137474987497280 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 11:19:43.696772 137474987497280 async_checkpointer.py:192] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7cf3f82374c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 11:19:44.835848 137474987497280 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_06_grad_accum/checkpoints I0422 11:19:45.029060 137474987497280 checkpoint_manager.py:929] [process=4][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_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7cf310409d60> I0422 11:19:45.029233 137474987497280 checkpointing.py:302] Checkpoint manager created! I0422 11:19:46.786859 137474987497280 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 11:19:46.786992 137474987497280 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0422 11:19:48.533826 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0422 11:19:48.533936 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0422 11:19:48.548913 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0422 11:19:48.548973 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0422 11:19:48.578550 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0422 11:19:48.578612 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0422 11:19:48.593495 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0422 11:19:48.593554 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0422 11:19:48.623260 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0422 11:19:48.623333 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0422 11:19:48.638219 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0422 11:19:48.638275 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0422 11:19:48.682661 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0422 11:19:48.682731 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0422 11:19:48.697773 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0422 11:19:48.697835 137474987497280 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0422 11:19:48.919725 137474987497280 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 11:19:48.919894 137474987497280 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 11:19:48.926301 137474987497280 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 11:19:48.926415 137474987497280 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 11:19:48.943038 137474987497280 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 11:19:48.943181 137474987497280 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 11:19:48.959278 137474987497280 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 11:19:48.959426 137474987497280 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 11:19:48.989503 137474987497280 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 11:19:48.989645 137474987497280 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 11:19:49.005747 137474987497280 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 11:19:49.005897 137474987497280 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 11:19:49.021934 137474987497280 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 11:19:49.022070 137474987497280 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 11:19:49.050006 137474987497280 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 11:19:49.050156 137474987497280 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 11:19:49.071393 137474987497280 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 11:19:49.071475 137474987497280 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 11:20:00.631118 137474987497280 max_utils.py:791] Total memory size: 1.7 GB, Output size: 0.4 GB, Temp size: 1.3 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0422 11:20:00.633457 137474987497280 metric_logger.py:301] number parameters: 1.104 billion I0422 11:20:02.268378 137474987497280 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0422 11:20:13.790345 137474987497280 checkpointing.py:798] Waited 11.521949291229248 seconds for step 0 to finish before starting checkpointing. I0422 11:20:13.792892 137474987497280 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 11:20:13.794451 137474987497280 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 0 I0422 11:20:13.795768 137474987497280 event_tracking.py:70] [process=4] [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_06_grad_accum/checkpoints/0. I0422 11:20:14.560028 137474987497280 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 11:20:14.561051 137474987497280 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 11:20:14.561109 137474987497280 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 11:20:14.833345 137474987497280 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274105s I0422 11:20:14.833532 137474987497280 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 5.445 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28328418731689453 s) (per-host) I0422 11:20:14.833602 137474987497280 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283366s (batch_requests_ready=0.003076s, total_serialization_initiated=0.280197s, others=0.000093s) I0422 11:20:14.833738 137474987497280 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287564s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.287545) I0422 11:20:14.834610 137474987497280 event_tracking.py:125] [process=4] [async] Finished blocking save in 1.04 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_06_grad_accum/checkpoints/0. I0422 11:20:14.834926 137349228918528 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 11:40:14.834891 I0422 11:20:14.844821 137474987497280 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 11:20:14.845094 137347030488832 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0422 11:20:14.845212 137474987497280 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_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776856813.7928548, 'wait_for_prev_duration_secs': 7.62939453125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776856813.7944899, 'checkpointer_blocking_duration_secs': 1.0405852794647217, 'get_old_steps_start_time': 1776856814.8351026, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776856813.7907453, 'checkpoint_manager_blocking_duration_secs': 1.0544285774230957} I0422 11:20:14.845321 137474987497280 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0422 11:20:14.845393 137474987497280 max_utils.py:750] Memstats: After params initialized: I0422 11:20:14.845442 137474987497280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0)) I0422 11:20:14.845476 137474987497280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0)) I0422 11:20:14.845503 137474987497280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0)) I0422 11:20:14.845527 137474987497280 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0)) I0422 11:20:15.163770 137474987497280 metric_logger.py:196] completed step: 0, seconds: 1.634, TFLOP/s/device: 33.264, Tokens/s/device: 5013.918, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0422 11:20:15.765426 137474987497280 metric_logger.py:196] completed step: 1, seconds: 12.894, TFLOP/s/device: 4.215, Tokens/s/device: 635.354, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0422 11:20:16.343950 137474987497280 metric_logger.py:196] completed step: 2, seconds: 0.025, TFLOP/s/device: 2217.221, Tokens/s/device: 334203.655, total_weights: 262144, loss: 10.561, lm_loss: 10.561, perplexity: 38601.980 I0422 11:20:16.922344 137474987497280 metric_logger.py:196] completed step: 3, seconds: 0.585, TFLOP/s/device: 92.966, Tokens/s/device: 14012.904, total_weights: 262144, loss: 10.270, lm_loss: 10.270, perplexity: 28865.557 I0422 11:20:17.797716 2565 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 11:20:18.079928 137474987497280 metric_logger.py:196] completed step: 4, seconds: 0.579, TFLOP/s/device: 93.876, Tokens/s/device: 14150.023, total_weights: 262144, loss: 10.021, lm_loss: 10.021, perplexity: 22497.775 I0422 11:20:18.087612 137474987497280 metric_logger.py:196] completed step: 5, seconds: 0.578, TFLOP/s/device: 94.050, Tokens/s/device: 14176.297, total_weights: 262144, loss: 9.819, lm_loss: 9.819, perplexity: 18383.324 I0422 11:20:19.817183 137347038881536 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 69 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_06_grad_accum/checkpoints/0/items/array_metadatas/process_4 I0422 11:20:52.856756 137349228918528 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 41.236 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.306477069854736 s) (per-host) I0422 11:20:52.856887 137349228918528 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.021850s. I0422 11:21:00.708100 137349228918528 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 45.873043s. I0422 11:21:00.708414 137347030488832 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 11:21:00.708529 137347030488832 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0422 11:21:00.708581 137347030488832 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 11:21:00.710937 137347030488832 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 11:21:03.842741 137474987497280 metric_logger.py:196] completed step: 6, seconds: 1.158, TFLOP/s/device: 46.916, Tokens/s/device: 7071.744, total_weights: 262144, loss: 9.666, lm_loss: 9.666, perplexity: 15775.117 I0422 11:21:04.421156 137474987497280 metric_logger.py:196] completed step: 7, seconds: 45.178, TFLOP/s/device: 1.203, Tokens/s/device: 181.327, total_weights: 262144, loss: 9.560, lm_loss: 9.560, perplexity: 14192.143 I0422 11:21:04.999489 137474987497280 metric_logger.py:196] completed step: 8, seconds: 0.585, TFLOP/s/device: 92.969, Tokens/s/device: 14013.240, total_weights: 262144, loss: 9.495, lm_loss: 9.495, perplexity: 13292.714 I0422 11:21:05.578494 137474987497280 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0422 11:21:05.579162 137474987497280 checkpointing.py:798] Waited 0.0006818771362304688 seconds for step 9 to finish before starting checkpointing. I0422 11:21:05.581171 137474987497280 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 11:21:05.582741 137474987497280 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 9 I0422 11:21:05.584188 137474987497280 event_tracking.py:70] [process=4] [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_06_grad_accum/checkpoints/9. I0422 11:21:06.305921 137474987497280 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 11:21:06.306015 137474987497280 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 11:21:06.339965 137474987497280 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035755s I0422 11:21:06.340131 137474987497280 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 36.178 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04263901710510254 s) (per-host) I0422 11:21:06.340184 137474987497280 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.042705s (batch_requests_ready=0.002468s, total_serialization_initiated=0.040163s, others=0.000073s) I0422 11:21:06.340280 137474987497280 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.046947s (all_items=0.000016s, per_item={'items': '0.00001621'}, temp_paths=0.046931) I0422 11:21:06.341044 137474987497280 event_tracking.py:125] [process=4] [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_06_grad_accum/checkpoints/9. I0422 11:21:06.341337 137342786467584 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 11:41:06.341305 I0422 11:21:06.343301 137474987497280 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 11:21:06.343561 137349192591104 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0422 11:21:06.343698 137474987497280 standard_logger.py:34] {'step': 9, '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_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776856865.5811386, 'wait_for_prev_duration_secs': 7.152557373046875e-05, 'time_between_consecutive_saves_sec': 4.870161771774292, 'checkpointer_blocking_start_time': 1776856865.5827787, 'checkpointer_blocking_duration_secs': 0.7587008476257324, 'get_old_steps_start_time': 1776856866.3415031, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776856865.579388, 'checkpoint_manager_blocking_duration_secs': 0.7642762660980225} I0422 11:21:06.343865 137474987497280 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0422 11:21:06.343925 137474987497280 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 11:21:11.277129 137349200983808 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 69 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_06_grad_accum/checkpoints/9/items/array_metadatas/process_4 I0422 11:21:48.419785 137342786467584 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.500 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.12225818634033 s) (per-host) I0422 11:21:48.419923 137342786467584 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.078481s. I0422 11:21:57.642718 137342786467584 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 51.301262s. I0422 11:21:57.643018 137349192591104 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 11:21:57.643134 137349192591104 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0422 11:21:57.643182 137349192591104 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 11:21:57.644631 137349192591104 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 11:21:57.644825 137474987497280 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0422 11:21:57.645024 137474987497280 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 11:21:57.646032 137474987497280 metric_logger.py:196] completed step: 9, seconds: 0.578, TFLOP/s/device: 94.054, Tokens/s/device: 14176.886, total_weights: 262144, loss: 9.457, lm_loss: 9.457, perplexity: 12793.173 Per train step: Total TFLOPs: 54.35 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Wed Apr 22 11:22:10 UTC 2026 EXIT_CODE=0