MaxView

XPK Start: Wed Apr 22 22:39:26 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. 
2026-04-22 22:39:51.574087: 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)
I0422 22:39:51.786717 139353305597760 max_utils.py:273] Attempting to initialize the jax distributed system...
I0422 22:40:00.826708 139353305597760 distributed.py:149] Starting JAX distributed service on [::]:8482
I0422 22:40:00.828983 139353305597760 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-t87ic-slice-job-0-0.mt-15-ocdbt-false-rs-t87ic:8482
I0422 22:40:02.133051 139353305597760 max_utils.py:284] Jax distributed system initialized!
I0422 22:40:08.353980 139353305597760 max_utils.py:800] System Information: Jax Version: 0.9.2
I0422 22:40:08.354089 139353305597760 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0422 22:40:08.354176 139353305597760 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
I0422 22:40:08.354214 139353305597760 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0422 22:40:09.049226 139353305597760 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0422 22:40:09.049515 139353305597760 checkpointing.py:677] Setting up checkpoint logger...
I0422 22:40:09.049569 139353305597760 checkpointing.py:233] Creating checkpoint manager with ocdbt=False and zarr3=True
I0422 22:40:09.049613 139353305597760 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0422 22:40:09.049951 139353305597760 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7ebd26d16f90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0422 22:40:10.360323 139353305597760 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0422 22:40:10.360562 139353305597760 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 0x7ea80472f950>}, handler_registry=None
I0422 22:40:10.360802 139353305597760 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ea80472f950>` 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`.
I0422 22:40:10.360849 139353305597760 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea804730bc0>` 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`.
I0422 22:40:10.360884 139353305597760 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 0x7ea80472f950>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ea80472f950>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea804730bc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea804730bc0>}).
I0422 22:40:10.361214 139353305597760 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0422 22:40:10.361290 139353305597760 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ea80456b560> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0422 22:40:13.500874 139231800309504 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776897480592897546, 'commit_timestamp_nsecs': 1776897494302408773, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/9/_CHECKPOINT_METADATA
I0422 22:40:13.747529 139231817094912 checkpoint_manager.py:1871] Missing metrics for step 0
I0422 22:40:14.584314 139353305597760 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints
I0422 22:40:14.925495 139353305597760 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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ea804730e00>
I0422 22:40:14.925663 139353305597760 checkpointing.py:301] Checkpoint manager created!
I0422 22:40:15.850989 139353305597760 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 22:40:15.851124 139353305597760 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 22:40:16.231365 139353305597760 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 22:40:16.231456 139353305597760 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 22:40:16.248058 139353305597760 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 22:40:16.248131 139353305597760 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 22:40:16.271855 139353305597760 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 22:40:16.271924 139353305597760 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 22:40:16.288693 139353305597760 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 22:40:16.288758 139353305597760 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 22:40:16.305384 139353305597760 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0422 22:40:16.305446 139353305597760 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 22:40:16.330451 139353305597760 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0422 22:40:16.330521 139353305597760 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 22:40:16.352491 139353305597760 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0422 22:40:16.352563 139353305597760 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0422 22:40:16.561999 139353305597760 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0422 22:40:16.562126 139353305597760 checkpointing.py:581] restoring from this run's directory step 9
I0422 22:40:16.805027 139353305597760 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/9.
I0422 22:40:17.057167 139353305597760 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/9.
I0422 22:40:17.057603 139353305597760 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/9.
I0422 22:40:20.136787    1489 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0422 22:40:29.676069 139353305597760 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 148.434 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.646285057067871 s) (per-host)
I0422 22:40:29.683140 139353305597760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 142.881 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.07634162902832 s) (per-host)
I0422 22:40:30.981958 139353305597760 event_tracking.py:138] [process=6] [sync] Finished load in 14.18 seconds @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/9
I0422 22:40:30.982142 139353305597760 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776897616.8049295, 'checkpointer_duration_secs': 14.17715835571289, 'checkpoint_manager_start_time': 1776897616.5623827, 'checkpoint_manager_duration_secs': 14.419707775115967}
fsdp: 32
I0422 22:40:31.401174 139353305597760 maxtext_utils.py:1668]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0422 22:40:31.401311 139353305597760 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)
I0422 22:40:31.401365 139353305597760 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)
I0422 22:40:31.401424 139353305597760 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')
I0422 22:40:31.401478 139353305597760 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)
I0422 22:40:31.401516 139353305597760 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)
I0422 22:40:31.401571 139353305597760 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)
I0422 22:40:31.401625 139353305597760 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')
I0422 22:40:31.401666 139353305597760 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)
I0422 22:40:31.401700 139353305597760 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)
I0422 22:40:31.401746 139353305597760 maxtext_utils.py:1668]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0422 22:40:31.401791 139353305597760 maxtext_utils.py:1668]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0422 22:40:31.897664 139353305597760 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0422 22:40:31.897758 139353305597760 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0422 22:40:31.913344 139353305597760 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0422 22:40:31.913406 139353305597760 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0422 22:40:42.963135 139353305597760 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.
I0422 22:40:42.963908 139353305597760 metric_logger.py:301] number parameters: 1.104 billion
I0422 22:40:54.880679 139353305597760 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0422 22:40:55.032843 139353305597760 checkpointing.py:776] Waited 0.15215206146240234 seconds for step 10 to finish before starting checkpointing.
I0422 22:40:55.035933 139353305597760 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 22:40:55.037665 139353305597760 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0422 22:40:55.038981 139353305597760 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/10.
I0422 22:40:56.217647 139353305597760 signaling_client.py:364] Using JaxDistributedSignalingClient
I0422 22:40:56.218625 139353305597760 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0422 22:40:56.218681 139353305597760 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
I0422 22:40:56.499586 139353305597760 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282013s
I0422 22:40:56.499754 139353305597760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.393 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28603076934814453 s) (per-host)
I0422 22:40:56.499806 139353305597760 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286092s (batch_requests_ready=0.002295s, total_serialization_initiated=0.283727s, others=0.000070s)
I0422 22:40:56.499895 139353305597760 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290118s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.290100)
I0422 22:40:56.500579 139353305597760 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.46 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/10.
I0422 22:40:56.500926 139228495189760 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 23:00:56.500887
I0422 22:40:56.515195 139353305597760 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 22:40:56.515495 139227975104256 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 22:40:56.515647 139353305597760 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776897655.0359142, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776897655.0377026, 'checkpointer_blocking_duration_secs': 1.463386058807373, 'get_old_steps_start_time': 1776897656.5011234, 'get_old_steps_duration_secs': 3.218650817871094e-05, 'checkpoint_manager_blocking_start_time': 1776897655.0333483, 'checkpoint_manager_blocking_duration_secs': 1.4822633266448975}
I0422 22:40:56.515754 139353305597760 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0422 22:40:56.515804 139353305597760 max_utils.py:750] 
Memstats: After params initialized:
I0422 22:40:56.515853 139353305597760 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0422 22:40:56.515884 139353305597760 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0422 22:40:56.515910 139353305597760 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0422 22:40:56.515934 139353305597760 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0422 22:40:56.828496 139353305597760 metric_logger.py:196] completed step: 10, seconds: 11.917, TFLOP/s/device: 1.140, Tokens/s/device: 171.860, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869
I0422 22:40:57.001703 139353305597760 metric_logger.py:196] completed step: 11, seconds: 1.946, TFLOP/s/device: 6.981, Tokens/s/device: 1052.275, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659
I0422 22:40:57.149000 139353305597760 metric_logger.py:196] completed step: 12, seconds: 0.026, TFLOP/s/device: 522.823, Tokens/s/device: 78805.603, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068
I0422 22:40:57.296519 139353305597760 metric_logger.py:196] completed step: 13, seconds: 0.154, TFLOP/s/device: 88.492, Tokens/s/device: 13338.544, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001
I0422 22:40:57.301399 139353305597760 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0422 22:40:57.590484 139353305597760 checkpointing.py:776] Waited 0.2890512943267822 seconds for step 15 to finish before starting checkpointing.
I0422 22:40:57.593417 139353305597760 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 22:41:01.827687 139227983496960 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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_6
I0422 22:41:07.497625 139228495189760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 139.987 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.283852815628052 s) (per-host)
I0422 22:41:07.497751 139228495189760 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.996708s.
I0422 22:41:14.441622 139228495189760 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.940561s.
I0422 22:41:14.441918 139227975104256 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 22:41:14.442039 139227975104256 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 22:41:14.442086 139227975104256 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0422 22:41:14.443430 139227975104256 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0422 22:41:14.443606 139353305597760 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.
W0422 22:41:14.443741 139353305597760 checkpoint_manager.py:1452] Waiting for previous save to complete took 16.850326 seconds. If this number is high, consider checkpointing less frequently.
I0422 22:41:14.445927 139353305597760 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15
I0422 22:41:14.448543 139353305597760 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/15.
I0422 22:41:15.170619 139353305597760 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0422 22:41:15.170712 139353305597760 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
I0422 22:41:15.196510 139353305597760 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026991s
I0422 22:41:15.196643 139353305597760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 49.479 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031176090240478516 s) (per-host)
I0422 22:41:15.196691 139353305597760 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031232s (batch_requests_ready=0.002527s, total_serialization_initiated=0.028644s, others=0.000062s)
I0422 22:41:15.196769 139353305597760 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.036340s (all_items=0.000013s, per_item={'items': '0.00001335'}, temp_paths=0.036327)
I0422 22:41:15.197469 139353305597760 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/15.
I0422 22:41:15.197779 139227471804160 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 23:01:15.197741
I0422 22:41:15.208192 139353305597760 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 22:41:15.208436 139227975104256 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 22:41:15.208571 139353305597760 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776897657.5933871, 'wait_for_prev_duration_secs': 16.85032558441162, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776897674.4459672, 'checkpointer_blocking_duration_secs': 0.7519509792327881, 'get_old_steps_start_time': 1776897675.197942, 'get_old_steps_duration_secs': 3.337860107421875e-05, 'checkpoint_manager_blocking_start_time': 1776897657.5911593, 'checkpoint_manager_blocking_duration_secs': 17.617374658584595}
I0422 22:41:15.208688 139353305597760 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0422 22:41:15.209595 139353305597760 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.291, Tokens/s/device: 13911.153, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193
I0422 22:41:15.215731 139353305597760 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 92.044, Tokens/s/device: 13873.835, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661
I0422 22:41:20.647207 139227983496960 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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_6
I0422 22:41:25.468718 139227471804160 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 153.311 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.303217887878418 s) (per-host)
I0422 22:41:25.468826 139227471804160 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.270935s.
I0422 22:41:33.488713 139227471804160 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 18.290807s.
I0422 22:41:33.488982 139227975104256 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 22:41:33.489129 139227975104256 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 22:41:33.489187 139227975104256 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0422 22:41:33.491022 139227975104256 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0422 22:41:36.178507 139353305597760 metric_logger.py:196] completed step: 16, seconds: 17.913, TFLOP/s/device: 0.759, Tokens/s/device: 114.332, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872
I0422 22:41:36.326012 139353305597760 metric_logger.py:196] completed step: 17, seconds: 20.817, TFLOP/s/device: 0.653, Tokens/s/device: 98.381, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975
I0422 22:41:36.473735 139353305597760 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.227, Tokens/s/device: 13449.263, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333
I0422 22:41:36.620799 139353305597760 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0422 22:41:36.621526 139353305597760 checkpointing.py:776] Waited 0.0007464885711669922 seconds for step 19 to finish before starting checkpointing.
I0422 22:41:36.623556 139353305597760 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 22:41:36.625225 139353305597760 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0422 22:41:36.626760 139353305597760 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/19.
I0422 22:41:37.324220 139353305597760 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0422 22:41:37.324319 139353305597760 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
I0422 22:41:37.348578 139353305597760 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.025475s
I0422 22:41:37.348712 139353305597760 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 53.051 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.02907729148864746 s) (per-host)
I0422 22:41:37.348767 139353305597760 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.029142s (batch_requests_ready=0.001796s, total_serialization_initiated=0.027276s, others=0.000070s)
I0422 22:41:37.348869 139353305597760 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.033248s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.033237)
I0422 22:41:37.349623 139353305597760 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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/19.
I0422 22:41:37.349943 139223688529664 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 23:01:37.349903
I0422 22:41:37.354958 139353305597760 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 22:41:37.355417 139227975104256 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0422 22:41:37.355577 139353305597760 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776897696.6235251, 'wait_for_prev_duration_secs': 6.985664367675781e-05, 'time_between_consecutive_saves_sec': 3.1324634552001953, 'checkpointer_blocking_start_time': 1776897696.6252627, 'checkpointer_blocking_duration_secs': 0.7248249053955078, 'get_old_steps_start_time': 1776897697.350119, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776897696.621756, 'checkpoint_manager_blocking_duration_secs': 0.7337877750396729}
I0422 22:41:37.355690 139353305597760 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0422 22:41:37.355737 139353305597760 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 22:41:42.403782 139227983496960 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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_6
I0422 22:41:47.761372 139223688529664 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 151.278 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.441699266433716 s) (per-host)
I0422 22:41:47.761494 139223688529664 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.411437s.
I0422 22:41:52.903708 139223688529664 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 15.553635s.
I0422 22:41:52.903989 139227975104256 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 22:41:52.904142 139227975104256 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0422 22:41:52.904186 139227975104256 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0422 22:41:52.906336 139227975104256 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0422 22:41:52.906527 139353305597760 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.
I0422 22:41:52.906681 139353305597760 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 22:41:52.907591 139353305597760 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.060, Tokens/s/device: 13876.279, 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: Wed Apr 22 22:42:01 UTC 2026
EXIT_CODE=0