XPK Start: Mon Apr 20 17:45:26 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 17:45:50.781584: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0420 17:45:50.959746 135059213608768 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 17:46:00.001091 135059213608768 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 17:46:00.003606 135059213608768 distributed.py:172] Connecting to JAX distributed service on mt-06-grad-accum-2koq9-slice-job-0-0.mt-06-grad-accum-2koq9:8482 I0420 17:46:00.928533 135059213608768 max_utils.py:284] Jax distributed system initialized! I0420 17:46:07.119602 135059213608768 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 17:46:07.119703 135059213608768 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 17:46:07.119746 135059213608768 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 I0420 17:46:07.119782 135059213608768 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 17:46:07.810786 135059213608768 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 17:46:07.923317 135059213608768 checkpointing.py:688] Setting up checkpoint logger... I0420 17:46:07.923431 135059213608768 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 17:46:07.923477 135059213608768 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 17:46:07.923678 135059213608768 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 0x7ad550918c80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 17:46:10.967302 135059213608768 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 17:46:10.967542 135059213608768 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ac100538dd0>}, handler_registry=None I0420 17:46:10.967777 135059213608768 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ac100538dd0>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0420 17:46:10.967824 135059213608768 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ac0e07473e0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0420 17:46:10.967860 135059213608768 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 0x7ac100538dd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ac100538dd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ac0e07473e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ac0e07473e0>}). I0420 17:46:10.968202 135059213608768 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 17:46:10.968273 135059213608768 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ac0e07ecfe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 17:46:12.450883 135059213608768 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints I0420 17:46:12.461444 135059213608768 checkpoint_manager.py:929] [process=5][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_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ac0e0749b80> I0420 17:46:12.461568 135059213608768 checkpointing.py:302] Checkpoint manager created! I0420 17:46:14.174341 135059213608768 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 17:46:14.174456 135059213608768 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 17:46:15.554302 135059213608768 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0420 17:46:15.554414 135059213608768 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:46:15.554466 135059213608768 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0420 17:46:15.554514 135059213608768 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0420 17:46:15.554551 135059213608768 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:46:15.554585 135059213608768 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0420 17:46:15.554620 135059213608768 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:46:15.554654 135059213608768 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0420 17:46:15.554687 135059213608768 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:46:15.554719 135059213608768 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0420 17:46:15.554751 135059213608768 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0420 17:46:15.554783 135059213608768 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0420 17:46:15.943633 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown. I0420 17:46:15.943731 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,). I0420 17:46:15.958616 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown. I0420 17:46:15.958674 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None). I0420 17:46:15.987823 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown. I0420 17:46:15.987886 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp'). I0420 17:46:16.002491 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown. I0420 17:46:16.002547 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None). I0420 17:46:16.031713 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown. I0420 17:46:16.031784 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None). I0420 17:46:16.046401 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown. I0420 17:46:16.046462 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp'). I0420 17:46:16.090326 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown. I0420 17:46:16.090406 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None). I0420 17:46:16.105034 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown. I0420 17:46:16.105090 135059213608768 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp'). I0420 17:46:16.323563 135059213608768 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:46:16.323715 135059213608768 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:46:16.330409 135059213608768 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:46:16.330513 135059213608768 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:46:16.347023 135059213608768 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:46:16.347152 135059213608768 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:46:16.363085 135059213608768 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:46:16.363213 135059213608768 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:46:16.393589 135059213608768 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:46:16.393733 135059213608768 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:46:16.409779 135059213608768 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:46:16.409907 135059213608768 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:46:16.425794 135059213608768 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:46:16.425918 135059213608768 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:46:16.454058 135059213608768 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 17:46:16.454197 135059213608768 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:46:16.475623 135059213608768 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 17:46:16.475730 135059213608768 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 17:46:28.045501 135059213608768 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. I0420 17:46:28.047932 135059213608768 metric_logger.py:301] number parameters: 1.104 billion I0420 17:46:29.689207 135059213608768 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 17:46:40.675292 135059213608768 checkpointing.py:798] Waited 10.98606824874878 seconds for step 0 to finish before starting checkpointing. I0420 17:46:40.677719 135059213608768 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:46:40.679369 135059213608768 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 0 I0420 17:46:40.680868 135059213608768 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/0. I0420 17:46:41.010963 135059213608768 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 17:46:41.011897 135059213608768 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 17:46:41.011970 135059213608768 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0420 17:46:41.285988 135059213608768 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.275790s I0420 17:46:41.286167 135059213608768 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 5.418 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2847278118133545 s) (per-host) I0420 17:46:41.286226 135059213608768 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284797s (batch_requests_ready=0.002939s, total_serialization_initiated=0.281781s, others=0.000077s) I0420 17:46:41.286320 135059213608768 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289048s (all_items=0.000019s, per_item={'items': '0.00001860'}, temp_paths=0.289030) I0420 17:46:41.287153 135059213608768 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/0. I0420 17:46:41.287435 134931058595584 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:06:41.287403 I0420 17:46:41.298533 135059213608768 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:46:41.298830 134930528986880 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:46:41.299009 135059213608768 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776707200.6776996, 'wait_for_prev_duration_secs': 6.031990051269531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776707200.6794071, 'checkpointer_blocking_duration_secs': 0.6081738471984863, 'get_old_steps_start_time': 1776707201.2876067, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776707200.6756787, 'checkpoint_manager_blocking_duration_secs': 0.6232657432556152} I0420 17:46:41.299118 135059213608768 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 17:46:41.299183 135059213608768 max_utils.py:750] Memstats: After params initialized: I0420 17:46:41.299234 135059213608768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0)) I0420 17:46:41.299266 135059213608768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0)) I0420 17:46:41.299294 135059213608768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0)) I0420 17:46:41.299319 135059213608768 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0)) I0420 17:46:41.612374 135059213608768 metric_logger.py:196] completed step: 0, seconds: 1.640, TFLOP/s/device: 33.136, Tokens/s/device: 4994.555, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0420 17:46:42.219147 135059213608768 metric_logger.py:196] completed step: 1, seconds: 11.921, TFLOP/s/device: 4.559, Tokens/s/device: 687.172, total_weights: 262144, loss: 10.876, lm_loss: 10.876, perplexity: 52883.598 I0420 17:46:42.797836 135059213608768 metric_logger.py:196] completed step: 2, seconds: 0.029, TFLOP/s/device: 1852.117, Tokens/s/device: 279171.210, total_weights: 262144, loss: 10.561, lm_loss: 10.561, perplexity: 38601.980 I0420 17:46:43.376582 135059213608768 metric_logger.py:196] completed step: 3, seconds: 0.585, TFLOP/s/device: 92.905, Tokens/s/device: 14003.682, total_weights: 262144, loss: 10.270, lm_loss: 10.270, perplexity: 28865.557 I0420 17:46:44.534788 135059213608768 metric_logger.py:196] completed step: 4, seconds: 0.579, TFLOP/s/device: 93.826, Tokens/s/device: 14142.401, total_weights: 262144, loss: 10.021, lm_loss: 10.021, perplexity: 22497.775 I0420 17:46:44.542313 135059213608768 metric_logger.py:196] completed step: 5, seconds: 0.578, TFLOP/s/device: 93.990, Tokens/s/device: 14167.201, total_weights: 262144, loss: 9.819, lm_loss: 9.819, perplexity: 18383.324 I0420 17:46:49.372979 2556 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 17:46:51.347727 134931041810176 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/0/items/array_metadatas/process_5 I0420 17:47:23.730313 134931058595584 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 36.968 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.7288339138031 s) (per-host) I0420 17:47:23.730437 134931058595584 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.442891s. I0420 17:47:30.057640 135059213608768 metric_logger.py:196] completed step: 6, seconds: 1.159, TFLOP/s/device: 46.899, Tokens/s/device: 7069.077, total_weights: 262144, loss: 9.666, lm_loss: 9.666, perplexity: 15775.117 I0420 17:47:30.636491 135059213608768 metric_logger.py:196] completed step: 7, seconds: 44.938, TFLOP/s/device: 1.209, Tokens/s/device: 182.297, total_weights: 262144, loss: 9.560, lm_loss: 9.560, perplexity: 14192.143 I0420 17:47:31.215585 135059213608768 metric_logger.py:196] completed step: 8, seconds: 0.585, TFLOP/s/device: 92.959, Tokens/s/device: 14011.850, total_weights: 262144, loss: 9.495, lm_loss: 9.495, perplexity: 13292.714 I0420 17:47:31.795183 135059213608768 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 17:47:31.795836 135059213608768 checkpointing.py:798] Waited 0.0006668567657470703 seconds for step 9 to finish before starting checkpointing. I0420 17:47:31.797922 135059213608768 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:47:32.280519 134931058595584 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 50.992957s. I0420 17:47:32.280802 134930528986880 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:47:32.280930 134930528986880 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:47:32.281002 134930528986880 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:47:32.284817 134930528986880 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 17:47:32.285012 135059213608768 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. I0420 17:47:32.286569 135059213608768 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 9 I0420 17:47:32.288506 135059213608768 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/9. I0420 17:47:32.583572 135059213608768 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 17:47:32.583668 135059213608768 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0420 17:47:32.622397 135059213608768 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040457s I0420 17:47:32.622532 135059213608768 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 32.588 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04733538627624512 s) (per-host) I0420 17:47:32.622579 135059213608768 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.047390s (batch_requests_ready=0.002606s, total_serialization_initiated=0.044723s, others=0.000061s) I0420 17:47:32.622658 135059213608768 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051467s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=0.051453) I0420 17:47:32.623342 135059213608768 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/9. I0420 17:47:32.625198 134930528986880 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:07:32.625175 I0420 17:47:32.627094 135059213608768 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:47:32.627317 134930543761152 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:47:32.627432 135059213608768 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776707251.7978923, 'wait_for_prev_duration_secs': 0.48723554611206055, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776707252.2866101, 'checkpointer_blocking_duration_secs': 0.33868956565856934, 'get_old_steps_start_time': 1776707252.6253216, 'get_old_steps_duration_secs': 2.7894973754882812e-05, 'checkpoint_manager_blocking_start_time': 1776707251.7960882, 'checkpoint_manager_blocking_duration_secs': 0.8313114643096924} I0420 17:47:32.627539 135059213608768 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 17:47:32.627582 135059213608768 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:47:38.325693 134934836532992 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_06_grad_accum/checkpoints/9/items/array_metadatas/process_5 I0420 17:48:14.915542 134930528986880 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.307 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.34031820297241 s) (per-host) I0420 17:48:14.915644 134930528986880 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.290373s. I0420 17:48:23.064051 134930528986880 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 50.438763s. I0420 17:48:23.064265 134930543761152 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:48:23.064344 134930543761152 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:48:23.064390 134930543761152 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:48:23.066804 134930543761152 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 17:48:23.066979 135059213608768 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 17:48:23.067170 135059213608768 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:48:23.068123 135059213608768 metric_logger.py:196] completed step: 9, seconds: 0.579, TFLOP/s/device: 93.845, Tokens/s/device: 14145.405, 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: Mon Apr 20 17:48:31 UTC 2026 EXIT_CODE=0