XPK Start: Fri Apr 24 14:34:45 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 14:35:10.376175: 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 14:35:10.950942 140534672512832 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:35:19.991423 140534672512832 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:35:19.996140 140534672512832 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-4u2jc-slice-job-0-0.mt-15-ocdbt-false-rs-4u2jc:8482 I0424 14:35:21.287868 140534672512832 max_utils.py:284] Jax distributed system initialized! I0424 14:35:27.400076 140534672512832 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:35:27.400202 140534672512832 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:35:27.400247 140534672512832 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 14:35:27.400282 140534672512832 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:35:28.101020 140534672512832 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:35:28.214498 140534672512832 checkpointing.py:688] Setting up checkpoint logger... I0424 14:35:28.214617 140534672512832 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0424 14:35:28.214663 140534672512832 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:35:28.214861 140534672512832 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7fd00fa3ce90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:35:29.486251 140534672512832 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:35:29.486594 140534672512832 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 0x7fbc4c197470>}, handler_registry=None I0424 14:35:29.486838 140534672512832 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fbc4c197470>` 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 14:35:29.486888 140534672512832 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fbc4c493ef0>` 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 14:35:29.486926 140534672512832 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 0x7fbc4c197470>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7fbc4c197470>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fbc4c493ef0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7fbc4c493ef0>}). I0424 14:35:29.487287 140534672512832 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:35:29.487359 140534672512832 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7fbc143df1a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:35:31.345937 140413444794112 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777041176731256895, 'commit_timestamp_nsecs': 1777041190853134489, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/5/_CHECKPOINT_METADATA I0424 14:35:31.613227 140413436401408 checkpoint_manager.py:1871] Missing metrics for step 9 I0424 14:35:31.885938 140534672512832 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints I0424 14:35:32.008723 140534672512832 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=5), 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7fbc14325520> I0424 14:35:32.008892 140534672512832 checkpointing.py:302] Checkpoint manager created! I0424 14:35:32.972832 140534672512832 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:35:32.972951 140534672512832 checkpointing.py:582] restoring from this run's directory step 9 I0424 14:35:33.228448 140534672512832 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9. I0424 14:35:33.467284 140534672512832 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9. I0424 14:35:33.467717 140534672512832 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.24 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9. I0424 14:35:36.520917 1480 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:35:43.791355 140534672512832 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 182.020 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 8.681881189346313 s) (per-host) I0424 14:35:43.813965 140534672512832 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 173.796 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 9.10611343383789 s) (per-host) I0424 14:35:45.526829 140534672512832 event_tracking.py:138] [process=6] [sync] Finished load in 12.30 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9 I0424 14:35:45.526984 140534672512832 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1777041333.2283518, 'checkpointer_duration_secs': 12.298601150512695, 'checkpoint_manager_start_time': 1777041332.9737484, 'checkpoint_manager_duration_secs': 12.553206443786621} fsdp: 32 I0424 14:35:45.603184 140534672512832 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:35:45.603297 140534672512832 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:35:45.603348 140534672512832 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:35:45.603390 140534672512832 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:35:45.603427 140534672512832 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:35:45.603460 140534672512832 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:35:45.603495 140534672512832 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:35:45.603533 140534672512832 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:35:45.603567 140534672512832 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:35:45.603599 140534672512832 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:35:45.603631 140534672512832 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:35:45.603662 140534672512832 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:35:46.008703 140534672512832 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:35:46.008801 140534672512832 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:35:46.014320 140534672512832 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:35:46.014378 140534672512832 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:35:46.030886 140534672512832 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:35:46.030948 140534672512832 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:35:46.046762 140534672512832 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:35:46.046820 140534672512832 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:35:46.070924 140534672512832 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 14:35:46.070995 140534672512832 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:35:46.087024 140534672512832 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 14:35:46.087084 140534672512832 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:35:46.103115 140534672512832 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 14:35:46.103178 140534672512832 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:35:46.133154 140534672512832 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:35:46.133225 140534672512832 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:35:46.153191 140534672512832 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:35:46.153261 140534672512832 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:35:57.804364 140534672512832 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 14:35:57.806679 140534672512832 metric_logger.py:301] number parameters: 1.104 billion I0424 14:36:09.450645 140534672512832 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 14:36:09.602378 140534672512832 checkpointing.py:798] Waited 0.1517176628112793 seconds for step 10 to finish before starting checkpointing. I0424 14:36:09.604751 140534672512832 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:36:09.606795 140534672512832 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0424 14:36:09.608395 140534672512832 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/10. I0424 14:36:09.937882 140534672512832 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:36:09.938804 140534672512832 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:36:09.938863 140534672512832 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 14:36:10.214265 140534672512832 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277183s I0424 14:36:10.214439 140534672512832 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.432 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28399133682250977 s) (per-host) I0424 14:36:10.214491 140534672512832 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284055s (batch_requests_ready=0.002476s, total_serialization_initiated=0.281506s, others=0.000072s) I0424 14:36:10.214599 140534672512832 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288280s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.288262) I0424 14:36:10.216803 140534672512832 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/10. I0424 14:36:10.217160 140410257139456 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:56:10.217121 I0424 14:36:10.224861 140534672512832 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:36:10.225175 140408605050624 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:36:10.225325 140534672512832 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041369.6047325, 'wait_for_prev_duration_secs': 6.270408630371094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041369.606835, 'checkpointer_blocking_duration_secs': 0.6104824542999268, 'get_old_steps_start_time': 1777041370.2173426, 'get_old_steps_duration_secs': 3.266334533691406e-05, 'checkpoint_manager_blocking_start_time': 1777041369.6027503, 'checkpoint_manager_blocking_duration_secs': 0.6225388050079346} I0424 14:36:10.225434 140534672512832 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 14:36:10.225508 140534672512832 max_utils.py:750] Memstats: After params initialized: I0424 14:36:10.225559 140534672512832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 14:36:10.225592 140534672512832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 14:36:10.225621 140534672512832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 14:36:10.225646 140534672512832 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 14:36:10.551265 140534672512832 metric_logger.py:196] completed step: 10, seconds: 11.643, TFLOP/s/device: 1.167, Tokens/s/device: 175.902, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0424 14:36:10.731293 140534672512832 metric_logger.py:196] completed step: 11, seconds: 1.099, TFLOP/s/device: 12.366, Tokens/s/device: 1863.962, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0424 14:36:10.878868 140534672512832 metric_logger.py:196] completed step: 12, seconds: 0.023, TFLOP/s/device: 581.915, Tokens/s/device: 87712.536, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0424 14:36:11.026884 140534672512832 metric_logger.py:196] completed step: 13, seconds: 0.165, TFLOP/s/device: 82.477, Tokens/s/device: 12431.862, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0424 14:36:11.033111 140534672512832 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 14:36:11.321555 140534672512832 checkpointing.py:798] Waited 0.2884242534637451 seconds for step 15 to finish before starting checkpointing. I0424 14:36:11.324294 140534672512832 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:36:15.109070 140409211557632 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_6 I0424 14:36:20.252657 140410257139456 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 153.029 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.322185754776001 s) (per-host) I0424 14:36:20.252774 140410257139456 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.035497s. I0424 14:36:26.876511 140410257139456 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 16.659217s. I0424 14:36:26.876798 140408605050624 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:36:26.876917 140408605050624 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:36:26.876967 140408605050624 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:36:26.878710 140408605050624 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 14:36:26.878874 140534672512832 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 14:36:26.879003 140534672512832 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.554709 seconds. If this number is high, consider checkpointing less frequently. I0424 14:36:26.880952 140534672512832 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0424 14:36:26.883444 140534672512832 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/15. I0424 14:36:27.172197 140534672512832 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:36:27.172296 140534672512832 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 14:36:27.198973 140534672512832 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028374s I0424 14:36:27.199173 140534672512832 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 43.280 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03564190864562988 s) (per-host) I0424 14:36:27.199228 140534672512832 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035708s (batch_requests_ready=0.002945s, total_serialization_initiated=0.032688s, others=0.000075s) I0424 14:36:27.199330 140534672512832 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.040628s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=0.040616) I0424 14:36:27.200007 140534672512832 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.32 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/15. I0424 14:36:27.200348 140408605050624 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:56:27.200310 I0424 14:36:27.218942 140534672512832 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:36:27.219211 140408596657920 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:36:27.219361 140534672512832 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041371.3242648, 'wait_for_prev_duration_secs': 15.554709434509277, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041386.8809907, 'checkpointer_blocking_duration_secs': 0.3195018768310547, 'get_old_steps_start_time': 1777041387.2005157, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1777041371.3219159, 'checkpoint_manager_blocking_duration_secs': 15.897410869598389} I0424 14:36:27.219473 140534672512832 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 14:36:27.220526 140534672512832 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.355, Tokens/s/device: 13920.704, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0424 14:36:27.228159 140534672512832 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.643, Tokens/s/device: 13813.478, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0424 14:36:32.482704 140409211557632 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_6 I0424 14:36:37.455599 140408605050624 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 153.478 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.292030811309814 s) (per-host) I0424 14:36:37.455726 140408605050624 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.255262s. I0424 14:36:43.297597 140408605050624 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 16.097118s. I0424 14:36:43.297874 140408596657920 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:36:43.297993 140408596657920 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:36:43.298044 140408596657920 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:36:43.299544 140408596657920 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 14:36:48.892425 140534672512832 metric_logger.py:196] completed step: 16, seconds: 16.193, TFLOP/s/device: 0.839, Tokens/s/device: 126.472, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0424 14:36:49.040156 140534672512832 metric_logger.py:196] completed step: 17, seconds: 21.519, TFLOP/s/device: 0.631, Tokens/s/device: 95.171, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0424 14:36:49.187823 140534672512832 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.696, Tokens/s/device: 13369.193, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0424 14:36:49.336250 140534672512832 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 14:36:49.336912 140534672512832 checkpointing.py:798] Waited 0.0006756782531738281 seconds for step 19 to finish before starting checkpointing. I0424 14:36:49.339012 140534672512832 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:36:49.340786 140534672512832 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0424 14:36:49.342178 140534672512832 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/19. I0424 14:36:49.649350 140534672512832 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:36:49.649497 140534672512832 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 14:36:49.675386 140534672512832 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027620s I0424 14:36:49.675508 140534672512832 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 44.601 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03458595275878906 s) (per-host) I0424 14:36:49.675552 140534672512832 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.034640s (batch_requests_ready=0.002656s, total_serialization_initiated=0.031926s, others=0.000058s) I0424 14:36:49.675624 140534672512832 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038788s (all_items=0.000011s, per_item={'items': '0.00001097'}, temp_paths=0.038777) I0424 14:36:49.676331 140534672512832 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/19. I0424 14:36:49.676666 140408596657920 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:56:49.676629 I0424 14:36:49.679714 140534672512832 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:36:49.679944 140405425301248 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:36:49.680067 140534672512832 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041409.3389819, 'wait_for_prev_duration_secs': 7.128715515136719e-05, 'time_between_consecutive_saves_sec': 6.039398908615112, 'checkpointer_blocking_start_time': 1777041409.3408241, 'checkpointer_blocking_duration_secs': 0.3359856605529785, 'get_old_steps_start_time': 1777041409.6768284, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1777041409.3371656, 'checkpoint_manager_blocking_duration_secs': 0.3428688049316406} I0424 14:36:49.680201 140534672512832 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 14:36:49.680254 140534672512832 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:36:55.220948 140409211557632 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_6 I0424 14:37:00.071938 140408596657920 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 151.433 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.430989503860474 s) (per-host) I0424 14:37:00.072044 140408596657920 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.395264s. I0424 14:37:06.802381 140408596657920 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.125585s. I0424 14:37:06.802663 140405425301248 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:37:06.802780 140405425301248 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:37:06.802821 140405425301248 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:37:06.804465 140405425301248 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 14:37:06.804651 140534672512832 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 14:37:06.804816 140534672512832 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:37:06.805786 140534672512832 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.752, Tokens/s/device: 13829.895, total_weights: 65536, loss: 5.385, lm_loss: 5.385, perplexity: 218.045 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 14:37:15 UTC 2026 EXIT_CODE=0