XPK Start: Sat Apr 25 21:26:53 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. 2026-04-25 21:27:18.719246: 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) I0425 21:27:18.928837 132392294319936 max_utils.py:273] Attempting to initialize the jax distributed system... I0425 21:27:27.970944 132392294319936 distributed.py:149] Starting JAX distributed service on [::]:8482 I0425 21:27:27.973350 132392294319936 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-rs-5sx66-slice-job-0-0.mt-16-zarr3-true-rs-5sx66:8482 I0425 21:27:28.913938 132392294319936 max_utils.py:284] Jax distributed system initialized! I0425 21:27:34.087417 132392294319936 max_utils.py:800] System Information: Jax Version: 0.9.2 I0425 21:27:34.087521 132392294319936 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0425 21:27:34.087560 132392294319936 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 I0425 21:27:34.087596 132392294319936 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 21:27:34.780911 132392294319936 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 21:27:34.781211 132392294319936 checkpointing.py:677] Setting up checkpoint logger... I0425 21:27:34.781271 132392294319936 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0425 21:27:34.781317 132392294319936 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0425 21:27:34.781665 132392294319936 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 0x786844238b60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 21:27:36.033984 132392294319936 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0425 21:27:36.034245 132392294319936 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 0x7854c4657c50>}, handler_registry=None I0425 21:27:36.034488 132392294319936 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7854c4657c50>` 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`. I0425 21:27:36.034537 132392294319936 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7854c4658920>` 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`. I0425 21:27:36.034574 132392294319936 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 0x7854c4657c50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7854c4657c50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7854c4658920>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7854c4658920>}). I0425 21:27:36.034940 132392294319936 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0425 21:27:36.035013 132392294319936 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7854ac27b2e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0425 21:27:37.982481 132273810679552 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777152292733959499, 'commit_timestamp_nsecs': 1777152337882348230, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/9/_CHECKPOINT_METADATA I0425 21:27:38.235122 132273810679552 checkpoint_manager.py:1871] Missing metrics for step 9 I0425 21:27:38.978739 132392294319936 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints I0425 21:27:39.541793 132392294319936 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/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7854c47bd130> I0425 21:27:39.541963 132392294319936 checkpointing.py:301] Checkpoint manager created! I0425 21:27:40.479365 132392294319936 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 21:27:40.479489 132392294319936 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 21:27:40.864477 132392294319936 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 21:27:40.864577 132392294319936 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 21:27:40.881120 132392294319936 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 21:27:40.881188 132392294319936 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 21:27:40.904911 132392294319936 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 21:27:40.904978 132392294319936 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 21:27:40.921653 132392294319936 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 21:27:40.921713 132392294319936 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 21:27:40.938180 132392294319936 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 21:27:40.938244 132392294319936 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 21:27:40.963205 132392294319936 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0425 21:27:40.963275 132392294319936 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 21:27:40.984320 132392294319936 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 21:27:40.984390 132392294319936 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 21:27:41.196481 132392294319936 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0425 21:27:41.196597 132392294319936 checkpointing.py:581] restoring from this run's directory step 9 I0425 21:27:41.432461 132392294319936 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/9. I0425 21:27:42.115873 132392294319936 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/9. I0425 21:27:42.116339 132392294319936 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.68 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/9. I0425 21:27:43.518003 1409 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 21:27:49.452062 132392294319936 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 266.035 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 5.940085411071777 s) (per-host) I0425 21:27:49.456830 132392294319936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 249.138 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.3522820472717285 s) (per-host) I0425 21:27:51.268886 132392294319936 event_tracking.py:138] [process=6] [sync] Finished load in 9.84 seconds @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/9 I0425 21:27:51.269043 132392294319936 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints', 'checkpointer_start_time': 1777152461.4323628, 'checkpointer_duration_secs': 9.836649656295776, 'checkpoint_manager_start_time': 1777152461.196844, 'checkpoint_manager_duration_secs': 10.072170495986938} fsdp: 32 I0425 21:27:51.687011 132392294319936 maxtext_utils.py:1668] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0425 21:27:51.687152 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 21:27:51.687207 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 21:27:51.687273 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0425 21:27:51.687326 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 21:27:51.687364 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 21:27:51.687414 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0425 21:27:51.687467 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0425 21:27:51.687508 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0425 21:27:51.687543 132392294319936 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0425 21:27:51.687587 132392294319936 maxtext_utils.py:1668] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0425 21:27:51.687630 132392294319936 maxtext_utils.py:1668] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0425 21:27:52.185032 132392294319936 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 21:27:52.185136 132392294319936 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0425 21:27:52.200637 132392294319936 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 21:27:52.200696 132392294319936 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0425 21:28:03.284872 132392294319936 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. I0425 21:28:03.285751 132392294319936 metric_logger.py:301] number parameters: 1.104 billion I0425 21:28:14.626886 132392294319936 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0425 21:28:14.809072 132392294319936 checkpointing.py:776] Waited 0.18216538429260254 seconds for step 10 to finish before starting checkpointing. I0425 21:28:14.811762 132392294319936 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 21:28:14.813898 132392294319936 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0425 21:28:14.815403 132392294319936 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/10. I0425 21:28:15.560614 132392294319936 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 21:28:15.561687 132392294319936 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 21:28:15.561745 132392294319936 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 I0425 21:28:15.838201 132392294319936 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.277671s I0425 21:28:15.838378 132392294319936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.478 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28157901763916016 s) (per-host) I0425 21:28:15.838433 132392294319936 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.281645s (batch_requests_ready=0.002174s, total_serialization_initiated=0.279397s, others=0.000075s) I0425 21:28:15.838528 132392294319936 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285807s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.285789) I0425 21:28:15.839304 132392294319936 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/10. I0425 21:28:15.839660 132273223485184 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 21:48:15.839621 I0425 21:28:15.844466 132392294319936 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0425 21:28:15.844765 132268899153664 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 21:28:15.844915 132392294319936 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777152494.8117433, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777152494.8139353, 'checkpointer_blocking_duration_secs': 1.02587890625, 'get_old_steps_start_time': 1777152495.8398383, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1777152494.809576, 'checkpoint_manager_blocking_duration_secs': 1.0353062152862549} I0425 21:28:15.845022 132392294319936 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0425 21:28:15.845071 132392294319936 max_utils.py:750] Memstats: After params initialized: I0425 21:28:15.845134 132392294319936 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0425 21:28:15.845166 132392294319936 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0425 21:28:15.845192 132392294319936 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0425 21:28:15.845216 132392294319936 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0425 21:28:16.160254 132392294319936 metric_logger.py:196] completed step: 10, seconds: 11.341, TFLOP/s/device: 1.198, Tokens/s/device: 180.583, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0425 21:28:16.764169 132392294319936 metric_logger.py:196] completed step: 11, seconds: 1.532, TFLOP/s/device: 8.871, Tokens/s/device: 1337.069, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0425 21:28:16.911739 132392294319936 metric_logger.py:196] completed step: 12, seconds: 0.457, TFLOP/s/device: 29.734, Tokens/s/device: 4481.812, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0425 21:28:17.058943 132392294319936 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.774, Tokens/s/device: 13380.986, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0425 21:28:17.063482 132392294319936 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0425 21:28:17.353156 132392294319936 checkpointing.py:776] Waited 0.28960490226745605 seconds for step 15 to finish before starting checkpointing. I0425 21:28:17.356126 132392294319936 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 21:28:21.156530 132268925462272 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/10/items/array_metadatas/process_6 I0425 21:28:56.987557 132273223485184 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.126 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.430739641189575 s) (per-host) I0425 21:28:56.987662 132273223485184 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.147892s. I0425 21:29:05.797985 132273223485184 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.958196s. I0425 21:29:05.798292 132268899153664 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 21:29:05.798410 132268899153664 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 21:29:05.798460 132268899153664 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0425 21:29:05.800010 132268899153664 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0425 21:29:05.800197 132392294319936 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. W0425 21:29:05.800326 132392294319936 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.444220 seconds. If this number is high, consider checkpointing less frequently. I0425 21:29:05.802380 132392294319936 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0425 21:29:05.805040 132392294319936 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/15. I0425 21:29:06.498041 132392294319936 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 21:29:06.498149 132392294319936 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 I0425 21:29:06.524180 132392294319936 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027304s I0425 21:29:06.524307 132392294319936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 50.290 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030673503875732422 s) (per-host) I0425 21:29:06.524353 132392294319936 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030728s (batch_requests_ready=0.001777s, total_serialization_initiated=0.028890s, others=0.000061s) I0425 21:29:06.524434 132392294319936 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035984s (all_items=0.000013s, per_item={'items': '0.00001264'}, temp_paths=0.035971) I0425 21:29:06.525110 132392294319936 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.72 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/15. I0425 21:29:06.525446 132267306415872 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 21:49:06.525411 I0425 21:29:06.538427 132392294319936 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0425 21:29:06.538717 132268899153664 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 21:29:06.538875 132392294319936 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777152497.3560781, 'wait_for_prev_duration_secs': 48.44422006607056, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777152545.8024192, 'checkpointer_blocking_duration_secs': 0.7231676578521729, 'get_old_steps_start_time': 1777152546.525611, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1777152497.3537536, 'checkpoint_manager_blocking_duration_secs': 49.18508863449097} I0425 21:29:06.538990 132392294319936 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0425 21:29:06.539905 132392294319936 metric_logger.py:196] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.069, Tokens/s/device: 13877.595, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0425 21:29:06.546046 132392294319936 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.311, Tokens/s/device: 13914.083, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0425 21:29:13.264522 132268925462272 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/15/items/array_metadatas/process_6 I0425 21:29:27.565417 132392294319936 metric_logger.py:196] completed step: 16, seconds: 49.481, TFLOP/s/device: 0.275, Tokens/s/device: 41.390, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0425 21:29:27.712946 132392294319936 metric_logger.py:196] completed step: 17, seconds: 20.873, TFLOP/s/device: 0.651, Tokens/s/device: 98.116, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0425 21:29:27.860201 132392294319936 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.238, Tokens/s/device: 13450.942, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0425 21:29:28.006876 132392294319936 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0425 21:29:28.007638 132392294319936 checkpointing.py:776] Waited 0.0007808208465576172 seconds for step 19 to finish before starting checkpointing. I0425 21:29:28.009703 132392294319936 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 21:29:48.830166 132267306415872 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.311 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.336475133895874 s) (per-host) I0425 21:29:48.830289 132267306415872 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.304729s. I0425 21:29:55.165487 132267306415872 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.639912s. I0425 21:29:55.165773 132268899153664 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 21:29:55.165889 132268899153664 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 21:29:55.165939 132268899153664 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0425 21:29:55.168124 132268899153664 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0425 21:29:55.168318 132392294319936 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0425 21:29:55.168446 132392294319936 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.158748 seconds. If this number is high, consider checkpointing less frequently. I0425 21:29:55.169880 132392294319936 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0425 21:29:55.172067 132392294319936 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/19. I0425 21:29:55.875202 132392294319936 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 21:29:55.875290 132392294319936 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 I0425 21:29:55.901267 132392294319936 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026911s I0425 21:29:55.901438 132392294319936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 50.843 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030340194702148438 s) (per-host) I0425 21:29:55.901489 132392294319936 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030401s (batch_requests_ready=0.001637s, total_serialization_initiated=0.028698s, others=0.000065s) I0425 21:29:55.901590 132392294319936 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034618s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.034606) I0425 21:29:55.902345 132392294319936 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/19. I0425 21:29:55.902640 132268925462272 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 21:49:55.902607 I0425 21:29:55.905498 132392294319936 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0425 21:29:55.905751 132268899153664 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 21:29:55.905875 132392294319936 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777152568.0096724, 'wait_for_prev_duration_secs': 27.158747673034668, 'time_between_consecutive_saves_sec': 22.209621906280518, 'checkpointer_blocking_start_time': 1777152595.1699183, 'checkpointer_blocking_duration_secs': 0.7328598499298096, 'get_old_steps_start_time': 1777152595.9027967, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1777152568.007879, 'checkpoint_manager_blocking_duration_secs': 27.8979651927948} I0425 21:29:55.905985 132392294319936 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0425 21:29:55.906027 132392294319936 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 21:30:01.499767 132267837990656 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_16_zarr3_true/checkpoints/19/items/array_metadatas/process_6 I0425 21:30:39.289007 132268925462272 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.381 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.417869091033936 s) (per-host) I0425 21:30:39.289154 132268925462272 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.386403s. I0425 21:30:45.397039 132268925462272 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.494272s. I0425 21:30:45.397352 132268899153664 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 21:30:45.397502 132268899153664 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 21:30:45.397638 132268899153664 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0425 21:30:45.399849 132268899153664 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0425 21:30:45.400026 132392294319936 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. I0425 21:30:45.400198 132392294319936 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 21:30:45.401205 132392294319936 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.229, Tokens/s/device: 13901.805, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sat Apr 25 21:30:53 UTC 2026 EXIT_CODE=0