XPK Start: Sat Apr 25 20:36:04 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 20:36:29.653637: 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 20:36:29.863770 135443237685056 max_utils.py:273] Attempting to initialize the jax distributed system... I0425 20:36:38.905152 135443237685056 distributed.py:149] Starting JAX distributed service on [::]:8482 I0425 20:36:38.907565 135443237685056 distributed.py:172] Connecting to JAX distributed service on mt-08-checkpoint-async--0ksb6-slice-job-0-0.mt-08-checkpoint-async--0ksb6:8482 I0425 20:36:40.061503 135443237685056 max_utils.py:284] Jax distributed system initialized! I0425 20:36:46.030805 135443237685056 max_utils.py:800] System Information: Jax Version: 0.9.2 I0425 20:36:46.030910 135443237685056 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0425 20:36:46.030950 135443237685056 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 20:36:46.030986 135443237685056 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 20:36:46.722538 135443237685056 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 20:36:46.722832 135443237685056 checkpointing.py:677] Setting up checkpoint logger... I0425 20:36:46.722887 135443237685056 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0425 20:36:46.722932 135443237685056 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0425 20:36:46.723266 135443237685056 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 0x7b2ec9919070>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 20:36:48.385983 135443237685056 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0425 20:36:48.386222 135443237685056 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1a1c23ebd0>}, handler_registry=None I0425 20:36:48.386467 135443237685056 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1a1c23ebd0>` 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 20:36:48.386515 135443237685056 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1a1c2df830>` 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 20:36:48.386552 135443237685056 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 0x7b1a1c23ebd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b1a1c23ebd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1a1c2df830>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b1a1c2df830>}). I0425 20:36:48.386895 135443237685056 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0425 20:36:48.386975 135443237685056 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b1a1c1072e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0425 20:36:50.296846 135320527959808 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777149145191237784, 'commit_timestamp_nsecs': 1777149187628225143, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA I0425 20:36:50.545607 135320527959808 checkpoint_manager.py:1871] Missing metrics for step 0 I0425 20:36:50.995751 135443237685056 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_08_checkpoint_async_true/checkpoints I0425 20:36:50.998228 135443237685056 checkpoint_manager.py:929] [process=5][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b2d922780b0> I0425 20:36:50.998342 135443237685056 checkpointing.py:301] Checkpoint manager created! I0425 20:36:51.935565 135443237685056 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 20:36:51.935700 135443237685056 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 20:36:52.321976 135443237685056 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 20:36:52.322076 135443237685056 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 20:36:52.338644 135443237685056 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0425 20:36:52.338725 135443237685056 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 20:36:52.362644 135443237685056 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 20:36:52.362721 135443237685056 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 20:36:52.379266 135443237685056 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 20:36:52.379327 135443237685056 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 20:36:52.395690 135443237685056 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 20:36:52.395750 135443237685056 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 20:36:52.420790 135443237685056 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0425 20:36:52.420867 135443237685056 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 20:36:52.441953 135443237685056 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 20:36:52.442019 135443237685056 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 20:36:52.651350 135443237685056 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0425 20:36:52.651464 135443237685056 checkpointing.py:581] restoring from this run's directory step 9 I0425 20:36:52.913175 135443237685056 event_tracking.py:70] [process=5] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/9. I0425 20:36:53.165426 135443237685056 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_08_checkpoint_async_true/checkpoints/9. I0425 20:36:53.165862 135443237685056 event_tracking.py:125] [process=5] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/9. I0425 20:36:54.566426 1466 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 20:37:00.562597 135443237685056 jax_array_handlers.py:843] [process=5] /jax/orbax/read/worker/io/requested throughput: 263.293 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.001964092254639 s) (per-host) I0425 20:37:00.567841 135443237685056 base_pytree_checkpoint_handler.py:130] [process=5] /jax/checkpoint/read/gbytes_per_sec: 247.381 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.397385597229004 s) (per-host) I0425 20:37:01.699806 135443237685056 event_tracking.py:138] [process=5] [sync] Finished load in 8.79 seconds @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/9 I0425 20:37:01.699958 135443237685056 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_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1777149412.913078, 'checkpointer_duration_secs': 8.786849021911621, 'checkpoint_manager_start_time': 1777149412.6517262, 'checkpoint_manager_duration_secs': 9.048202514648438} fsdp: 32 I0425 20:37:02.119141 135443237685056 maxtext_utils.py:1668] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0425 20:37:02.119278 135443237685056 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 20:37:02.119333 135443237685056 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 20:37:02.119392 135443237685056 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 20:37:02.119444 135443237685056 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 20:37:02.119481 135443237685056 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 20:37:02.119533 135443237685056 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 20:37:02.119584 135443237685056 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 20:37:02.119628 135443237685056 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 20:37:02.119676 135443237685056 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 20:37:02.119726 135443237685056 maxtext_utils.py:1668] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0425 20:37:02.119772 135443237685056 maxtext_utils.py:1668] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0425 20:37:02.607702 135443237685056 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 20:37:02.607792 135443237685056 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0425 20:37:02.623253 135443237685056 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 20:37:02.623318 135443237685056 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0425 20:37:13.579065 135443237685056 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 20:37:13.579876 135443237685056 metric_logger.py:301] number parameters: 1.104 billion I0425 20:37:24.862306 135443237685056 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0425 20:37:25.201613 135443237685056 checkpointing.py:776] Waited 0.33928537368774414 seconds for step 10 to finish before starting checkpointing. I0425 20:37:25.204586 135443237685056 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 20:37:25.206444 135443237685056 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 10 I0425 20:37:25.207934 135443237685056 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/10. I0425 20:37:25.945757 135443237685056 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 20:37:25.946792 135443237685056 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 20:37:25.946847 135443237685056 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 20:37:26.218082 135443237685056 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272402s I0425 20:37:26.218248 135443237685056 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 5.584 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2762448787689209 s) (per-host) I0425 20:37:26.218305 135443237685056 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.276312s (batch_requests_ready=0.002137s, total_serialization_initiated=0.274100s, others=0.000076s) I0425 20:37:26.218392 135443237685056 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.280334s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.280316) I0425 20:37:26.219126 135443237685056 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/10. I0425 20:37:26.219476 135320477603584 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 20:57:26.219440 I0425 20:37:26.228665 135443237685056 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0425 20:37:26.228941 135318363690752 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0425 20:37:26.229098 135443237685056 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777149445.2045684, 'wait_for_prev_duration_secs': 5.91278076171875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777149445.206482, 'checkpointer_blocking_duration_secs': 1.0131475925445557, 'get_old_steps_start_time': 1777149446.2196658, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1777149445.2021735, 'checkpoint_manager_blocking_duration_secs': 1.0268902778625488} I0425 20:37:26.229199 135443237685056 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0425 20:37:26.229252 135443237685056 max_utils.py:750] Memstats: After params initialized: I0425 20:37:26.229300 135443237685056 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0)) I0425 20:37:26.229331 135443237685056 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0)) I0425 20:37:26.229355 135443237685056 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0)) I0425 20:37:26.229378 135443237685056 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0)) I0425 20:37:26.544864 135443237685056 metric_logger.py:196] completed step: 10, seconds: 11.282, TFLOP/s/device: 1.204, Tokens/s/device: 181.523, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0425 20:37:26.719603 135443237685056 metric_logger.py:196] completed step: 11, seconds: 1.681, TFLOP/s/device: 8.083, Tokens/s/device: 1218.311, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0425 20:37:26.866983 135443237685056 metric_logger.py:196] completed step: 12, seconds: 0.028, TFLOP/s/device: 491.610, Tokens/s/device: 74100.876, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0425 20:37:27.014192 135443237685056 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.693, Tokens/s/device: 13368.844, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0425 20:37:27.018718 135443237685056 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0425 20:37:27.308366 135443237685056 checkpointing.py:776] Waited 0.2896158695220947 seconds for step 15 to finish before starting checkpointing. I0425 20:37:27.311315 135443237685056 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 20:37:32.457115 135318372083456 array_metadata_store.py:203] [process=5][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_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_5 I0425 20:38:08.437076 135320477603584 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.171 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.49502658843994 s) (per-host) I0425 20:38:08.437203 135320477603584 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.217614s. I0425 20:38:16.310745 135320477603584 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 50.091138s. I0425 20:38:16.310988 135318363690752 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 20:38:16.311056 135318363690752 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0425 20:38:16.311108 135318363690752 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0425 20:38:16.312998 135318363690752 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0425 20:38:16.313191 135443237685056 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0425 20:38:16.313334 135443237685056 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.002018 seconds. If this number is high, consider checkpointing less frequently. I0425 20:38:16.315561 135443237685056 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 15 I0425 20:38:16.317996 135443237685056 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/15. I0425 20:38:16.601424 135443237685056 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 20:38:16.601517 135443237685056 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 20:38:16.625965 135443237685056 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.025614s I0425 20:38:16.626093 135443237685056 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 53.269 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.02895808219909668 s) (per-host) I0425 20:38:16.626141 135443237685056 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.029014s (batch_requests_ready=0.001741s, total_serialization_initiated=0.027212s, others=0.000061s) I0425 20:38:16.626216 135443237685056 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034062s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.034049) I0425 20:38:16.626910 135443237685056 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.31 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/15. I0425 20:38:16.627261 135318363690752 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 20:58:16.627224 I0425 20:38:16.640532 135443237685056 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0425 20:38:16.640822 135315105023744 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0425 20:38:16.640982 135443237685056 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777149447.3112845, 'wait_for_prev_duration_secs': 49.002018451690674, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777149496.3156, 'checkpointer_blocking_duration_secs': 0.3118016719818115, 'get_old_steps_start_time': 1777149496.6274247, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1777149447.308971, 'checkpoint_manager_blocking_duration_secs': 49.33197736740112} I0425 20:38:16.641099 135443237685056 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0425 20:38:16.641918 135443237685056 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.237, Tokens/s/device: 13902.937, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0425 20:38:16.648030 135443237685056 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.346, Tokens/s/device: 13919.379, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0425 20:38:21.952150 135318372083456 array_metadata_store.py:203] [process=5][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_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_5 I0425 20:38:37.883836 135443237685056 metric_logger.py:196] completed step: 16, seconds: 49.628, TFLOP/s/device: 0.274, Tokens/s/device: 41.267, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0425 20:38:38.031527 135443237685056 metric_logger.py:196] completed step: 17, seconds: 21.090, TFLOP/s/device: 0.644, Tokens/s/device: 97.107, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0425 20:38:38.178613 135443237685056 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.433, Tokens/s/device: 13480.336, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0425 20:38:38.325369 135443237685056 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0425 20:38:38.326092 135443237685056 checkpointing.py:776] Waited 0.0007419586181640625 seconds for step 19 to finish before starting checkpointing. I0425 20:38:38.328141 135443237685056 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 20:38:59.215811 135318363690752 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.064 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.61863565444946 s) (per-host) I0425 20:38:59.215934 135318363690752 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.588561s. I0425 20:39:07.090240 135318363690752 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 50.462842s. I0425 20:39:07.090525 135315105023744 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 20:39:07.090643 135315105023744 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0425 20:39:07.090705 135315105023744 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0425 20:39:07.091975 135315105023744 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0425 20:39:07.092149 135443237685056 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0425 20:39:07.092291 135443237685056 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.764146 seconds. If this number is high, consider checkpointing less frequently. I0425 20:39:07.093815 135443237685056 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 19 I0425 20:39:07.095917 135443237685056 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260425_201227/linen_xpk_test_pipeline_scan_nnx_20260425_201227_08_checkpoint_async_true/checkpoints/19. I0425 20:39:07.793905 135443237685056 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 20:39:07.793995 135443237685056 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 20:39:07.820769 135443237685056 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027726s I0425 20:39:07.820936 135443237685056 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 49.553 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031130075454711914 s) (per-host) I0425 20:39:07.820988 135443237685056 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031192s (batch_requests_ready=0.001604s, total_serialization_initiated=0.029521s, others=0.000067s) I0425 20:39:07.821082 135443237685056 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035331s (all_items=0.000011s, per_item={'items': '0.00001097'}, temp_paths=0.035320) I0425 20:39:07.821760 135443237685056 event_tracking.py:125] [process=5] [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_08_checkpoint_async_true/checkpoints/19. I0425 20:39:07.822071 135319403861760 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 20:59:07.822034 I0425 20:39:07.824169 135443237685056 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0425 20:39:07.824398 135315105023744 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0425 20:39:07.824532 135443237685056 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_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777149518.32811, 'wait_for_prev_duration_secs': 28.764145612716675, 'time_between_consecutive_saves_sec': 22.015073776245117, 'checkpointer_blocking_start_time': 1777149547.0938542, 'checkpointer_blocking_duration_secs': 0.728355884552002, 'get_old_steps_start_time': 1777149547.8222258, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1777149518.3263257, 'checkpoint_manager_blocking_duration_secs': 29.49817395210266} I0425 20:39:07.824636 135443237685056 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0425 20:39:07.824693 135443237685056 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 20:39:12.921133 135318372083456 array_metadata_store.py:203] [process=5][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_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_5 I0425 20:39:49.578491 135319403861760 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.800 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.78862428665161 s) (per-host) I0425 20:39:49.578617 135319403861760 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.756437s. I0425 20:39:57.961455 135319403861760 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 50.139258s. I0425 20:39:57.961776 135315105023744 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 20:39:57.961904 135315105023744 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0425 20:39:57.961947 135315105023744 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0425 20:39:57.963412 135315105023744 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0425 20:39:57.963570 135443237685056 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0425 20:39:57.963752 135443237685056 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 20:39:57.964931 135443237685056 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.032, Tokens/s/device: 13872.049, 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 20:40:10 UTC 2026 EXIT_CODE=0