XPK Start: Sun Apr 19 01:57:36 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 01:58:00.890429: 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) I0419 01:58:01.066682 132830815225664 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 01:58:10.106487 132830815225664 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 01:58:10.108829 132830815225664 distributed.py:172] Connecting to JAX distributed service on mt-11-optimizer-offload-rmwyx-slice-job-0-0.mt-11-optimizer-offload-rmwyx:8482 I0419 01:58:11.926072 132830815225664 max_utils.py:284] Jax distributed system initialized! I0419 01:58:18.020737 132830815225664 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 01:58:18.020835 132830815225664 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 01:58:18.020873 132830815225664 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 I0419 01:58:18.020906 132830815225664 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 01:58:19.077073 132830815225664 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 01:58:19.195748 132830815225664 checkpointing.py:688] Setting up checkpoint logger... I0419 01:58:19.195870 132830815225664 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 01:58:19.195914 132830815225664 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 01:58:19.196118 132830815225664 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 0x78ce89118440>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 01:58:22.199908 132830815225664 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 01:58:22.200148 132830815225664 checkpoint_manager.py:708] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78ba603f88c0>}, handler_registry=None I0419 01:58:22.200403 132830815225664 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78ba603f88c0>` 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`. I0419 01:58:22.200453 132830815225664 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78cd522825d0>` 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`. I0419 01:58:22.200489 132830815225664 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 0x78ba603f88c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78ba603f88c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78cd522825d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78cd522825d0>}). I0419 01:58:22.200815 132830815225664 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 01:58:22.200891 132830815225664 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78ba6015c4a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 01:58:24.074610 132830815225664 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints I0419 01:58:24.078369 132830815225664 checkpoint_manager.py:929] [process=3][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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78cd5228a1e0> I0419 01:58:24.078482 132830815225664 checkpointing.py:302] Checkpoint manager created! I0419 01:58:25.049131 132830815225664 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 01:58:25.049259 132830815225664 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 01:58:26.414615 132830815225664 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 01:58:26.414732 132830815225664 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:58:26.414786 132830815225664 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 01:58:26.414829 132830815225664 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 01:58:26.414865 132830815225664 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:58:26.414899 132830815225664 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 01:58:26.414934 132830815225664 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:58:26.414970 132830815225664 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 01:58:26.415003 132830815225664 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:58:26.415035 132830815225664 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 01:58:26.415066 132830815225664 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 01:58:26.415102 132830815225664 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 01:58:26.815067 132830815225664 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:58:26.815167 132830815225664 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:58:26.821734 132830815225664 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 01:58:26.821793 132830815225664 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:58:26.838194 132830815225664 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:58:26.838263 132830815225664 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:58:26.853880 132830815225664 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 01:58:26.853943 132830815225664 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 01:58:26.878433 132830815225664 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:58:26.878503 132830815225664 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:58:26.894171 132830815225664 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:58:26.894228 132830815225664 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:58:26.909871 132830815225664 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 01:58:26.909934 132830815225664 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:58:26.942367 132830815225664 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 01:58:26.942440 132830815225664 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 01:58:26.962680 132830815225664 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 01:58:26.962746 132830815225664 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 01:58:38.180063 132830815225664 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. I0419 01:58:38.182480 132830815225664 metric_logger.py:301] number parameters: 1.104 billion I0419 01:58:49.270206 132830815225664 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 01:58:49.422447 132830815225664 checkpointing.py:798] Waited 0.15222740173339844 seconds for step 0 to finish before starting checkpointing. I0419 01:58:49.424919 132830815225664 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 01:58:49.426591 132830815225664 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0419 01:58:49.428042 132830815225664 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/0. I0419 01:58:50.579820 132830815225664 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 01:58:50.580823 132830815225664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 01:58:50.580881 132830815225664 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 I0419 01:58:50.852486 132830815225664 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.273410s I0419 01:58:50.852661 132830815225664 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.459 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2825751304626465 s) (per-host) I0419 01:58:50.852715 132830815225664 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.282639s (batch_requests_ready=0.003216s, total_serialization_initiated=0.279351s, others=0.000071s) I0419 01:58:50.852819 132830815225664 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.286978s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.286961) I0419 01:58:50.853591 132830815225664 event_tracking.py:125] [process=3] [async] Finished blocking save in 1.43 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/0. I0419 01:58:50.853888 132702409848576 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:18:50.853854 I0419 01:58:50.864773 132830815225664 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:58:50.865061 132701826242304 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:58:50.865217 132830815225664 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776563929.4249, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776563929.4266298, 'checkpointer_blocking_duration_secs': 1.427396297454834, 'get_old_steps_start_time': 1776563930.8540504, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776563929.422834, 'checkpoint_manager_blocking_duration_secs': 1.4423456192016602} I0419 01:58:50.865341 132830815225664 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 01:58:50.865411 132830815225664 max_utils.py:750] Memstats: After params initialized: I0419 01:58:50.865463 132830815225664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0419 01:58:50.865498 132830815225664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0419 01:58:50.865522 132830815225664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0419 01:58:50.865545 132830815225664 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0419 01:58:51.183262 132830815225664 metric_logger.py:196] completed step: 0, seconds: 11.087, TFLOP/s/device: 1.226, Tokens/s/device: 184.726, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 01:58:51.368835 132830815225664 metric_logger.py:196] completed step: 1, seconds: 1.911, TFLOP/s/device: 7.110, Tokens/s/device: 1071.644, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 01:58:51.767769 132830815225664 metric_logger.py:196] completed step: 2, seconds: 0.039, TFLOP/s/device: 345.570, Tokens/s/device: 52088.102, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28632.133 I0419 01:58:51.915591 132830815225664 metric_logger.py:196] completed step: 3, seconds: 0.389, TFLOP/s/device: 34.946, Tokens/s/device: 5267.517, total_weights: 65536, loss: 9.732, lm_loss: 9.732, perplexity: 16855.789 I0419 01:58:52.211204 132830815225664 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.544, Tokens/s/device: 12441.982, total_weights: 65536, loss: 9.274, lm_loss: 9.274, perplexity: 10661.594 I0419 01:58:52.218875 132830815225664 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.216, Tokens/s/device: 13899.824, total_weights: 65536, loss: 8.888, lm_loss: 8.888, perplexity: 7245.978 I0419 01:58:55.579471 2816 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 01:58:58.369590 132701834635008 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_3 I0419 01:59:16.746969 132830815225664 metric_logger.py:196] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.845, Tokens/s/device: 6910.304, total_weights: 65536, loss: 8.589, lm_loss: 8.589, perplexity: 5371.156 I0419 01:59:16.894841 132830815225664 metric_logger.py:196] completed step: 7, seconds: 24.382, TFLOP/s/device: 0.557, Tokens/s/device: 83.998, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4360.465 I0419 01:59:17.042578 132830815225664 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.585, Tokens/s/device: 13352.458, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.256 I0419 01:59:17.190950 132830815225664 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 01:59:17.191632 132830815225664 checkpointing.py:798] Waited 0.00069427490234375 seconds for step 9 to finish before starting checkpointing. I0419 01:59:17.193669 132830815225664 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:59:29.220694 132702409848576 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 40.869 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.65054535865784 s) (per-host) I0419 01:59:29.220823 132702409848576 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.366833s. I0419 01:59:39.293764 132702409848576 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 48.439758s. I0419 01:59:39.294077 132701826242304 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 01:59:39.294209 132701826242304 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 01:59:39.294291 132701826242304 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 01:59:39.295904 132701826242304 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 01:59:39.296091 132830815225664 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 01:59:39.296259 132830815225664 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.102572 seconds. If this number is high, consider checkpointing less frequently. I0419 01:59:39.298162 132830815225664 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0419 01:59:39.300214 132830815225664 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/9. I0419 01:59:40.011211 132830815225664 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 01:59:40.011316 132830815225664 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 I0419 01:59:40.047144 132830815225664 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037531s I0419 01:59:40.047319 132830815225664 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 34.550 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.044647932052612305 s) (per-host) I0419 01:59:40.047371 132830815225664 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044711s (batch_requests_ready=0.002840s, total_serialization_initiated=0.041800s, others=0.000071s) I0419 01:59:40.047463 132830815225664 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049122s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.049107) I0419 01:59:40.048205 132830815225664 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/9. I0419 01:59:40.048547 132701826242304 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:19:40.048510 I0419 01:59:40.052133 132830815225664 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 01:59:40.052385 132680112920320 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0419 01:59:40.052511 132830815225664 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776563957.1936402, 'wait_for_prev_duration_secs': 22.102571725845337, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776563979.298201, 'checkpointer_blocking_duration_secs': 0.7504816055297852, 'get_old_steps_start_time': 1776563980.0487087, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776563957.1918607, 'checkpoint_manager_blocking_duration_secs': 22.860616207122803} I0419 01:59:40.052623 132830815225664 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 01:59:40.052668 132830815225664 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 01:59:48.084481 132701834635008 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_3 I0419 02:00:23.005534 132701826242304 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 36.732 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.0028190612793 s) (per-host) I0419 02:00:23.005656 132701826242304 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.957000s. I0419 02:00:30.906815 132701826242304 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.858145s. I0419 02:00:30.907108 132680112920320 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:00:30.907226 132680112920320 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:00:30.907286 132680112920320 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:00:30.908731 132680112920320 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 02:00:30.908897 132830815225664 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 02:00:30.909071 132830815225664 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:00:30.909922 132830815225664 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.581, Tokens/s/device: 13804.074, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3551.438 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 02:00:42 UTC 2026 EXIT_CODE=0