XPK Start: Fri Apr 24 14:26:16 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 14:26:41.861783: 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 14:26:42.077636 139258131679040 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:26:51.119070 139258131679040 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:26:51.121492 139258131679040 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-rs-ay59m-slice-job-0-0.mt-15-ocdbt-true-rs-ay59m:8482 I0424 14:26:51.521471 139258131679040 max_utils.py:284] Jax distributed system initialized! I0424 14:26:57.650237 139258131679040 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:26:57.650345 139258131679040 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:26:57.650385 139258131679040 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 14:26:57.650422 139258131679040 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:26:58.358685 139258131679040 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:26:58.473989 139258131679040 checkpointing.py:688] Setting up checkpoint logger... I0424 14:26:58.474111 139258131679040 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 14:26:58.474157 139258131679040 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:26:58.474355 139258131679040 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 0x7ea6ec718e60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:26:59.829785 139258131679040 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:26:59.830043 139258131679040 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 0x7e92b4142720>}, handler_registry=None I0424 14:26:59.830287 139258131679040 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e92b4142720>` 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 14:26:59.830338 139258131679040 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e92b40f9670>` 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 14:26:59.830375 139258131679040 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 0x7e92b4142720>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e92b4142720>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e92b40f9670>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e92b40f9670>}). I0424 14:26:59.830704 139258131679040 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:26:59.830784 139258131679040 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e92b40ff560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:27:01.511907 139136899983104 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777040598901648095, 'commit_timestamp_nsecs': 1777040646400531965, '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_true/checkpoints/5/_CHECKPOINT_METADATA I0424 14:27:01.759323 139136908375808 checkpoint_manager.py:1871] Missing metrics for step 0 I0424 14:27:02.107407 139258131679040 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_true/checkpoints I0424 14:27:02.584344 139258131679040 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=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_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e92b415d9d0> I0424 14:27:02.584532 139258131679040 checkpointing.py:302] Checkpoint manager created! I0424 14:27:03.553632 139258131679040 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:27:03.553749 139258131679040 checkpointing.py:582] restoring from this run's directory step 9 I0424 14:27:03.804061 139258131679040 event_tracking.py:70] [process=3] [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_true/checkpoints/9. I0424 14:27:04.052285 139258131679040 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_true/checkpoints/9. I0424 14:27:04.052690 139258131679040 event_tracking.py:125] [process=3] [sync] Finished blocking load in 0.25 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_true/checkpoints/9. I0424 14:27:05.813294 1453 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:27:12.459841 139258131679040 jax_array_handlers.py:843] [process=3] /jax/orbax/read/worker/io/requested throughput: 237.525 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.653097867965698 s) (per-host) I0424 14:27:12.469286 139258131679040 base_pytree_checkpoint_handler.py:130] [process=3] /jax/checkpoint/read/gbytes_per_sec: 223.213 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.090124845504761 s) (per-host) I0424 14:27:15.339575 139258131679040 event_tracking.py:138] [process=3] [sync] Finished load in 11.54 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_true/checkpoints/9 I0424 14:27:15.339728 139258131679040 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_true/checkpoints', 'checkpointer_start_time': 1777040823.8039622, 'checkpointer_duration_secs': 11.535732984542847, 'checkpoint_manager_start_time': 1777040823.5543664, 'checkpoint_manager_duration_secs': 11.785330533981323} fsdp: 32 I0424 14:27:15.416907 139258131679040 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:27:15.417033 139258131679040 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:27:15.417085 139258131679040 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:27:15.417127 139258131679040 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:27:15.417164 139258131679040 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:27:15.417199 139258131679040 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:27:15.417236 139258131679040 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:27:15.417271 139258131679040 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:27:15.417307 139258131679040 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:27:15.417341 139258131679040 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:27:15.417375 139258131679040 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:27:15.417408 139258131679040 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:27:15.825553 139258131679040 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:27:15.825649 139258131679040 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:27:15.831105 139258131679040 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:27:15.831164 139258131679040 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:27:15.848079 139258131679040 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:27:15.848140 139258131679040 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:27:15.864424 139258131679040 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:27:15.864484 139258131679040 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:27:15.889999 139258131679040 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 14:27:15.890067 139258131679040 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:27:15.906440 139258131679040 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 14:27:15.906502 139258131679040 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:27:15.922831 139258131679040 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 14:27:15.922892 139258131679040 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:27:15.952903 139258131679040 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:27:15.952984 139258131679040 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:27:15.973422 139258131679040 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:27:15.973484 139258131679040 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:27:27.292285 139258131679040 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 14:27:27.296066 139258131679040 metric_logger.py:301] number parameters: 1.104 billion I0424 14:27:38.467336 139258131679040 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 14:27:38.619743 139258131679040 checkpointing.py:798] Waited 0.15237188339233398 seconds for step 10 to finish before starting checkpointing. I0424 14:27:38.622021 139258131679040 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:27:38.623795 139258131679040 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 10 I0424 14:27:38.625333 139258131679040 event_tracking.py:70] [process=3] [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_true/checkpoints/10. I0424 14:27:38.959142 139258131679040 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:27:38.960087 139258131679040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:27:38.960144 139258131679040 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 14:27:39.238371 139258131679040 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280013s I0424 14:27:39.238539 139258131679040 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.376 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28695058822631836 s) (per-host) I0424 14:27:39.238593 139258131679040 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287015s (batch_requests_ready=0.002614s, total_serialization_initiated=0.284328s, others=0.000073s) I0424 14:27:39.238690 139258131679040 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290987s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.290969) I0424 14:27:39.239451 139258131679040 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.62 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_true/checkpoints/10. I0424 14:27:39.239789 139135801095936 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:47:39.239751 I0424 14:27:39.244357 139258131679040 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:27:39.244698 139131988457216 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:27:39.244851 139258131679040 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040858.622004, 'wait_for_prev_duration_secs': 5.793571472167969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777040858.6238353, 'checkpointer_blocking_duration_secs': 0.6161212921142578, 'get_old_steps_start_time': 1777040859.2399805, 'get_old_steps_duration_secs': 3.361701965332031e-05, 'checkpoint_manager_blocking_start_time': 1777040858.6201396, 'checkpoint_manager_blocking_duration_secs': 0.6246757507324219} I0424 14:27:39.244974 139258131679040 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 14:27:39.245046 139258131679040 max_utils.py:750] Memstats: After params initialized: I0424 14:27:39.245095 139258131679040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0424 14:27:39.245129 139258131679040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0424 14:27:39.245156 139258131679040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0424 14:27:39.245182 139258131679040 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0424 14:27:39.561966 139258131679040 metric_logger.py:196] completed step: 10, seconds: 11.170, TFLOP/s/device: 1.216, Tokens/s/device: 183.345, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537 I0424 14:27:39.750978 139258131679040 metric_logger.py:196] completed step: 11, seconds: 1.093, TFLOP/s/device: 12.433, Tokens/s/device: 1874.011, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821 I0424 14:27:39.898494 139258131679040 metric_logger.py:196] completed step: 12, seconds: 0.032, TFLOP/s/device: 421.502, Tokens/s/device: 63533.426, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013 I0424 14:27:40.046034 139258131679040 metric_logger.py:196] completed step: 13, seconds: 0.166, TFLOP/s/device: 81.943, Tokens/s/device: 12351.338, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520 I0424 14:27:40.052696 139258131679040 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 14:27:40.340867 139258131679040 checkpointing.py:798] Waited 0.28814125061035156 seconds for step 15 to finish before starting checkpointing. I0424 14:27:40.343692 139258131679040 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:27:44.449368 139132563085056 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_true/checkpoints/10/items/array_metadatas/process_3 I0424 14:28:20.777125 139135801095936 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.766 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.82550001144409 s) (per-host) I0424 14:28:20.777250 139135801095936 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.537348s. I0424 14:28:29.987033 139135801095936 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.747115s. I0424 14:28:29.987357 139131988457216 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:28:29.987475 139131988457216 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:28:29.987523 139131988457216 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:28:29.989053 139131988457216 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 14:28:29.989224 139258131679040 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0424 14:28:29.989360 139258131679040 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.645668 seconds. If this number is high, consider checkpointing less frequently. I0424 14:28:29.991464 139258131679040 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 15 I0424 14:28:29.994065 139258131679040 event_tracking.py:70] [process=3] [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_true/checkpoints/15. I0424 14:28:30.336399 139258131679040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:28:30.336494 139258131679040 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 14:28:30.361682 139258131679040 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026845s I0424 14:28:30.361853 139258131679040 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 45.619 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.033814430236816406 s) (per-host) I0424 14:28:30.361906 139258131679040 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.033877s (batch_requests_ready=0.002640s, total_serialization_initiated=0.031166s, others=0.000071s) I0424 14:28:30.362015 139258131679040 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.039610s (all_items=0.000035s, per_item={'items': '0.00003481'}, temp_paths=0.039575) I0424 14:28:30.363873 139258131679040 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.37 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_true/checkpoints/15. I0424 14:28:30.364206 139131988457216 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:48:30.364171 I0424 14:28:30.388114 139258131679040 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:28:30.388409 139132571477760 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:28:30.388571 139258131679040 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040860.3436627, 'wait_for_prev_duration_secs': 49.64566779136658, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777040909.9915042, 'checkpointer_blocking_duration_secs': 0.372861385345459, 'get_old_steps_start_time': 1777040910.364391, 'get_old_steps_duration_secs': 3.7670135498046875e-05, 'checkpoint_manager_blocking_start_time': 1777040860.34125, 'checkpoint_manager_blocking_duration_secs': 50.04728722572327} I0424 14:28:30.388682 139258131679040 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 14:28:30.389772 139258131679040 metric_logger.py:196] completed step: 14, seconds: 0.146, TFLOP/s/device: 92.936, Tokens/s/device: 14008.400, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067 I0424 14:28:30.397696 139258131679040 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.784, Tokens/s/device: 13834.660, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780 I0424 14:28:35.922986 139129316681472 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_true/checkpoints/15/items/array_metadatas/process_3 I0424 14:28:52.173373 139258131679040 metric_logger.py:196] completed step: 16, seconds: 50.343, TFLOP/s/device: 0.270, Tokens/s/device: 40.681, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201 I0424 14:28:52.321453 139258131679040 metric_logger.py:196] completed step: 17, seconds: 21.631, TFLOP/s/device: 0.628, Tokens/s/device: 94.681, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443 I0424 14:28:52.468856 139258131679040 metric_logger.py:196] completed step: 18, seconds: 0.154, TFLOP/s/device: 88.462, Tokens/s/device: 13334.028, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909 I0424 14:28:52.616878 139258131679040 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 14:28:52.617579 139258131679040 checkpointing.py:798] Waited 0.0007135868072509766 seconds for step 19 to finish before starting checkpointing. I0424 14:28:52.620161 139258131679040 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:29:12.766473 139131988457216 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.221 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.43839478492737 s) (per-host) I0424 14:29:12.766599 139131988457216 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.402280s. I0424 14:29:18.927633 139131988457216 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 48.563298s. I0424 14:29:18.927915 139132571477760 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:29:18.928051 139132571477760 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:29:18.928099 139132571477760 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:29:18.929531 139132571477760 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 14:29:18.929733 139258131679040 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0424 14:29:18.929865 139258131679040 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.309709 seconds. If this number is high, consider checkpointing less frequently. I0424 14:29:18.932195 139258131679040 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 19 I0424 14:29:18.934268 139258131679040 event_tracking.py:70] [process=3] [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_true/checkpoints/19. I0424 14:29:19.238194 139258131679040 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:29:19.238289 139258131679040 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 14:29:19.282394 139258131679040 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045820s I0424 14:29:19.282561 139258131679040 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 29.183 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.052858829498291016 s) (per-host) I0424 14:29:19.282612 139258131679040 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.052920s (batch_requests_ready=0.002746s, total_serialization_initiated=0.050106s, others=0.000068s) I0424 14:29:19.282715 139258131679040 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.056986s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.056974) I0424 14:29:19.283452 139258131679040 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.35 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_true/checkpoints/19. I0424 14:29:19.283749 139132571477760 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:49:19.283718 I0424 14:29:19.285710 139258131679040 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:29:19.285919 139129316681472 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:29:19.286052 139258131679040 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040932.6201324, 'wait_for_prev_duration_secs': 26.309709310531616, 'time_between_consecutive_saves_sec': 22.631040811538696, 'checkpointer_blocking_start_time': 1777040958.932234, 'checkpointer_blocking_duration_secs': 0.35165953636169434, 'get_old_steps_start_time': 1777040959.283911, 'get_old_steps_duration_secs': 4.553794860839844e-05, 'checkpoint_manager_blocking_start_time': 1777040932.6178195, 'checkpoint_manager_blocking_duration_secs': 26.66820240020752} I0424 14:29:19.286170 139258131679040 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 14:29:19.286212 139258131679040 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:29:24.861713 139130389608192 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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_true/checkpoints/19/items/array_metadatas/process_3 I0424 14:30:01.621155 139132571477760 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.262 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.39141225814819 s) (per-host) I0424 14:30:01.621284 139132571477760 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.337424s. I0424 14:30:08.243171 139132571477760 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 48.959296s. I0424 14:30:08.243390 139129316681472 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:30:08.243471 139129316681472 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:30:08.243514 139129316681472 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:30:08.245060 139129316681472 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 14:30:08.245232 139258131679040 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0424 14:30:08.245416 139258131679040 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:30:08.246375 139258131679040 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.577, Tokens/s/device: 13803.423, 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 14:30:21 UTC 2026 EXIT_CODE=0