MaxView

Timing Test Reference Updates
‹ 10_shardy_trueCase: 11_optimizer_offload_false11_optimizer_offload_true ›

Metrics: main (574ad3fb9) vs test/pipeline-scan-nnx (574ad3fb9)

Metricmain  574ad3fb9test/pipeline-scan-nnx  574ad3fb9Diff (test/pipeline-scan-nnx − main)
Parameters1.104 billion1.104 billion—
Final loss8.17908.17900
TFLOP/s92.22392.145-0.078
Tok/s13900.913889.1-11.784
Avg s/step2.9352.434-0.501
Memory %1.381.380
JAX0.9.20.8.1—

Diff = branch value − main value. Green = branch improved. Red = branch regressed.

main  ·  574ad3fb9  ·  main_20260418_180002  ·  full log 
XPK Start: Sat Apr 18 18:48:48 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-18 18:49:12.500909: 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)
I0418 18:49:13.035605 136439124154176 max_utils.py:273] Attempting to initialize the jax distributed system...
I0418 18:49:22.077880 136439124154176 distributed.py:149] Starting JAX distributed service on [::]:8482
I0418 18:49:22.080379 136439124154176 distributed.py:172] Connecting to JAX distributed service on mt-11-optimizer-offload-loa72-slice-job-0-0.mt-11-optimizer-offload-loa72:8482
I0418 18:49:23.826328 136439124154176 max_utils.py:284] Jax distributed system initialized!
I0418 18:49:28.905785 136439124154176 max_utils.py:800] System Information: Jax Version: 0.9.2
I0418 18:49:28.905889 136439124154176 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0418 18:49:28.905928 136439124154176 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
I0418 18:49:28.905961 136439124154176 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0418 18:49:29.605794 136439124154176 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0418 18:49:29.606091 136439124154176 checkpointing.py:677] Setting up checkpoint logger...
I0418 18:49:29.606146 136439124154176 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0418 18:49:29.606190 136439124154176 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0418 18:49:29.606526 136439124154176 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 0x7c1694b18620>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0418 18:49:32.917521 136439124154176 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0418 18:49:32.917747 136439124154176 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c02e02f2210>}, handler_registry=None
I0418 18:49:32.917975 136439124154176 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c02e02f2210>` 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`.
I0418 18:49:32.918021 136439124154176 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c02e00f8e60>` 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`.
I0418 18:49:32.918065 136439124154176 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 0x7c02e02f2210>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c02e02f2210>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c02e00f8e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c02e00f8e60>}).
I0418 18:49:32.918385 136439124154176 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0418 18:49:32.918452 136439124154176 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c01d048b4c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0418 18:49:34.494021 136439124154176 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints
I0418 18:49:35.789996 136439124154176 checkpoint_manager.py:929] [process=7][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=10), 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_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c157231b500>
I0418 18:49:35.790198 136439124154176 checkpointing.py:301] Checkpoint manager created!
I0418 18:49:36.741262 136439124154176 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0418 18:49:36.741375 136439124154176 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0418 18:49:37.115400 136439124154176 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0418 18:49:37.115497 136439124154176 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0418 18:49:37.131680 136439124154176 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0418 18:49:37.131742 136439124154176 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0418 18:49:37.154982 136439124154176 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0418 18:49:37.155051 136439124154176 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0418 18:49:37.171104 136439124154176 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0418 18:49:37.171162 136439124154176 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0418 18:49:37.187949 136439124154176 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0418 18:49:37.188007 136439124154176 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0418 18:49:37.213296 136439124154176 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0418 18:49:37.213365 136439124154176 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0418 18:49:37.234015 136439124154176 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0418 18:49:37.234099 136439124154176 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0418 18:49:37.448215 136439124154176 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0418 18:49:37.448318 136439124154176 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0418 18:49:38.859098 136439124154176 maxtext_utils.py:1654]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0418 18:49:38.859223 136439124154176 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0418 18:49:38.859275 136439124154176 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0418 18:49:38.859330 136439124154176 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   P('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0418 18:49:38.859400 136439124154176 maxtext_utils.py:1654]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0418 18:49:38.859464 136439124154176 maxtext_utils.py:1654]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0418 18:49:38.859519 136439124154176 maxtext_utils.py:1654]  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)
I0418 18:49:38.859568 136439124154176 maxtext_utils.py:1654]  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')
I0418 18:49:38.859605 136439124154176 maxtext_utils.py:1654]  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)
I0418 18:49:38.859640 136439124154176 maxtext_utils.py:1654]  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)
I0418 18:49:38.859684 136439124154176 maxtext_utils.py:1654]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0418 18:49:38.859729 136439124154176 maxtext_utils.py:1654]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0418 18:49:39.343242 136439124154176 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0418 18:49:39.343334 136439124154176 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0418 18:49:39.358338 136439124154176 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0418 18:49:39.358396 136439124154176 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0418 18:49:50.380766 136439124154176 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.
I0418 18:49:50.381576 136439124154176 metric_logger.py:301] number parameters: 1.104 billion
I0418 18:50:01.492863 136439124154176 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0418 18:50:01.839738 136439124154176 checkpointing.py:776] Waited 0.3468594551086426 seconds for step 0 to finish before starting checkpointing.
I0418 18:50:01.842085 136439124154176 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0418 18:50:01.844000 136439124154176 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0
I0418 18:50:01.845449 136439124154176 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/0.
I0418 18:50:02.592069 136439124154176 signaling_client.py:364] Using JaxDistributedSignalingClient
I0418 18:50:02.593109 136439124154176 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0418 18:50:02.593166 136439124154176 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
I0418 18:50:02.872663 136439124154176 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280666s
I0418 18:50:02.872834 136439124154176 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.391 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28613948822021484 s) (per-host)
I0418 18:50:02.872889 136439124154176 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286210s (batch_requests_ready=0.002231s, total_serialization_initiated=0.283900s, others=0.000078s)
I0418 18:50:02.872978 136439124154176 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290934s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.290917)
I0418 18:50:02.873782 136439124154176 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/0.
I0418 18:50:02.874124 136313551439616 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:10:02.874086
I0418 18:50:02.877295 136439124154176 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0418 18:50:02.877519 136310383703808 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save.
I0418 18:50:02.877640 136439124154176 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538201.8420558, 'wait_for_prev_duration_secs': 6.890296936035156e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538201.8440363, 'checkpointer_blocking_duration_secs': 1.030238389968872, 'get_old_steps_start_time': 1776538202.874299, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776538201.8402777, 'checkpoint_manager_blocking_duration_secs': 1.0373256206512451}
I0418 18:50:02.877750 136439124154176 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0418 18:50:02.877799 136439124154176 max_utils.py:750] 
Memstats: After params initialized:
I0418 18:50:02.877850 136439124154176 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0))
I0418 18:50:02.877881 136439124154176 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0))
I0418 18:50:02.877908 136439124154176 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0))
I0418 18:50:02.877932 136439124154176 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0))
I0418 18:50:03.190033 136439124154176 metric_logger.py:196] completed step: 0, seconds: 11.111, TFLOP/s/device: 1.223, Tokens/s/device: 184.319, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0418 18:50:03.363527 136439124154176 metric_logger.py:196] completed step: 1, seconds: 1.696, TFLOP/s/device: 8.013, Tokens/s/device: 1207.823, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0418 18:50:03.764384 136439124154176 metric_logger.py:196] completed step: 2, seconds: 0.026, TFLOP/s/device: 515.915, Tokens/s/device: 77764.277, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0418 18:50:03.911560 136439124154176 metric_logger.py:196] completed step: 3, seconds: 0.401, TFLOP/s/device: 33.879, Tokens/s/device: 5106.659, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0418 18:50:04.206683 136439124154176 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.891, Tokens/s/device: 13398.582, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0418 18:50:04.212482 136439124154176 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.281, Tokens/s/device: 13909.642, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0418 18:50:06.971169    2779 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0418 18:50:09.817960 136311983093504 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_7
I0418 18:50:27.759218 136439124154176 metric_logger.py:196] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.971, Tokens/s/device: 6929.243, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0418 18:50:27.906634 136439124154176 metric_logger.py:196] completed step: 7, seconds: 23.400, TFLOP/s/device: 0.581, Tokens/s/device: 87.521, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0418 18:50:28.054253 136439124154176 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.414, Tokens/s/device: 13477.497, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0418 18:50:28.200857 136439124154176 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0418 18:50:28.201534 136439124154176 checkpointing.py:776] Waited 0.0006995201110839844 seconds for step 9 to finish before starting checkpointing.
I0418 18:50:28.203721 136439124154176 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0418 18:50:40.373033 136313551439616 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 41.803 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.78630447387695 s) (per-host)
I0418 18:50:40.373172 136313551439616 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.498937s.
I0418 18:50:50.248514 136313551439616 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 47.374262s.
I0418 18:50:50.248794 136310383703808 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save.
I0418 18:50:50.248916 136310383703808 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save.
I0418 18:50:50.248966 136310383703808 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0418 18:50:50.250754 136310383703808 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0418 18:50:50.250927 136439124154176 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0418 18:50:50.251055 136439124154176 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.047334 seconds. If this number is high, consider checkpointing less frequently.
I0418 18:50:50.252955 136439124154176 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9
I0418 18:50:50.254985 136439124154176 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/9.
I0418 18:50:50.970291 136439124154176 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0418 18:50:50.970387 136439124154176 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
I0418 18:50:51.004894 136439124154176 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035756s
I0418 18:50:51.005060 136439124154176 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 39.381 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03917098045349121 s) (per-host)
I0418 18:50:51.005132 136439124154176 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039251s (batch_requests_ready=0.001762s, total_serialization_initiated=0.037401s, others=0.000088s)
I0418 18:50:51.005226 136439124154176 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043500s (all_items=0.000015s, per_item={'items': '0.00001550'}, temp_paths=0.043484)
I0418 18:50:51.005897 136439124154176 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/9.
I0418 18:50:51.006235 136310383703808 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:10:51.006199
I0418 18:50:51.008144 136439124154176 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0418 18:50:51.008375 136285712787200 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save.
I0418 18:50:51.008501 136439124154176 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538228.2036915, 'wait_for_prev_duration_secs': 22.04733443260193, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538250.2529948, 'checkpointer_blocking_duration_secs': 0.7533867359161377, 'get_old_steps_start_time': 1776538251.0064046, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776538228.2017593, 'checkpoint_manager_blocking_duration_secs': 22.806710958480835}
I0418 18:50:51.008662 136439124154176 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0418 18:50:51.008705 136439124154176 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0418 18:50:58.728660 136311983093504 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_7
I0418 18:51:33.357195 136310383703808 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.262 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.39127707481384 s) (per-host)
I0418 18:51:33.357300 136310383703808 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.350953s.
I0418 18:51:42.573403 136310383703808 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 51.567040s.
I0418 18:51:42.573677 136285712787200 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save.
I0418 18:51:42.573791 136285712787200 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save.
I0418 18:51:42.573837 136285712787200 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0418 18:51:42.575446 136285712787200 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0418 18:51:42.575612 136439124154176 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0418 18:51:42.575758 136439124154176 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0418 18:51:42.576667 136439124154176 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.223, Tokens/s/device: 13900.861, total_weights: 65536, loss: 8.179, lm_loss: 8.179, perplexity: 3564.635
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Sat Apr 18 18:51:53 UTC 2026
EXIT_CODE=0
test/pipeline-scan-nnx  ·  574ad3fb9  ·  test_pipeline_scan_nnx_20260419_085454  ·  full log 
XPK Start: Sun Apr 19 09:33:58 UTC 2026
2026-04-19 09:34:03.024090: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered
WARNING: All log messages before absl::InitializeLog() is called are written to STDERR
E0000 00:00:1776591243.036594      10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered
E0000 00:00:1776591243.040260      10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered
W0000 00:00:1776591243.051482      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776591243.051501      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776591243.051503      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
W0000 00:00:1776591243.051505      10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once.
2026-04-19 09:34:22.152750: 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)
I0419 09:34:22.673074 139115631470400 max_utils.py:273] Attempting to initialize the jax distributed system...
INFO:2026-04-19 09:34:31,713:jax._src.distributed:140: Starting JAX distributed service on [::]:8482
I0419 09:34:31.713533 139115631470400 distributed.py:140] Starting JAX distributed service on [::]:8482
INFO:2026-04-19 09:34:31,715:jax._src.distributed:157: Connecting to JAX distributed service on mt-11-optimizer-offload-6nhp6-slice-job-0-0.mt-11-optimizer-offload-6nhp6:8482
I0419 09:34:31.715930 139115631470400 distributed.py:157] Connecting to JAX distributed service on mt-11-optimizer-offload-6nhp6-slice-job-0-0.mt-11-optimizer-offload-6nhp6:8482
I0419 09:34:33.362840 139115631470400 max_utils.py:284] Jax distributed system initialized!
I0419 09:34:40.279614 139115631470400 max_utils.py:800] System Information: Jax Version: 0.8.1
I0419 09:34:40.279721 139115631470400 max_utils.py:801] System Information: Jaxlib Version: 0.8.1
I0419 09:34:40.279761 139115631470400 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709
I0419 09:34:40.279795 139115631470400 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0419 09:34:40.897491 139115631470400 maxtext_utils.py:1398] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0419 09:34:40.897785 139115631470400 checkpointing.py:677] Setting up checkpoint logger...
I0419 09:34:40.897841 139115631470400 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0419 09:34:40.897890 139115631470400 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None
I0419 09:34:40.898274 139115631470400 base_pytree_checkpoint_handler.py:415] 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 0x7e85d1571370>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0419 09:34:43.752367 139115631470400 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0419 09:34:43.752711 139115631470400 checkpoint_manager.py:709] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e849c9a1f40>}, handler_registry=None
I0419 09:34:43.752946 139115631470400 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e849c9a1f40>` 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`.
I0419 09:34:43.752995 139115631470400 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e734c2b7770>` 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`.
I0419 09:34:43.753032 139115631470400 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 0x7e849c9a1f40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e849c9a1f40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e734c2b7770>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e734c2b7770>}).
I0419 09:34:43.755681 139115631470400 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33
I0419 09:34:43.755765 139115631470400 async_checkpointer.py:177] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e85a9db72e0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0419 09:34:44.868583 139115631470400 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints
I0419 09:34:45.019377 139115631470400 checkpoint_manager.py:929] [process=3][thread=MainThread] CheckpointManager created,  primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), 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_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e734cc69a00>
I0419 09:34:45.019556 139115631470400 checkpointing.py:301] Checkpoint manager created!
I0419 09:34:45.910722 139115631470400 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0419 09:34:45.910829 139115631470400 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0419 09:34:46.282402 139115631470400 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0419 09:34:46.282499 139115631470400 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0419 09:34:46.298807 139115631470400 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0419 09:34:46.298867 139115631470400 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0419 09:34:46.322470 139115631470400 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0419 09:34:46.322540 139115631470400 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0419 09:34:46.338546 139115631470400 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0419 09:34:46.338608 139115631470400 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0419 09:34:46.354366 139115631470400 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0419 09:34:46.354428 139115631470400 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0419 09:34:46.378327 139115631470400 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0419 09:34:46.378395 139115631470400 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0419 09:34:46.397863 139115631470400 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0419 09:34:46.397927 139115631470400 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0419 09:34:46.592692 139115631470400 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0419 09:34:46.592811 139115631470400 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0419 09:34:48.018540 139115631470400 maxtext_utils.py:1501]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   PartitionSpec('norm',)
    Physical:  (None,)
I0419 09:34:48.018672 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0419 09:34:48.018724 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   PartitionSpec('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0419 09:34:48.018781 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   PartitionSpec('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0419 09:34:48.018832 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0419 09:34:48.018871 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   PartitionSpec('norm', 'layers')
    Physical:  (None, None)
I0419 09:34:48.018924 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0419 09:34:48.018977 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   PartitionSpec('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0419 09:34:48.019017 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0419 09:34:48.019054 139115631470400 maxtext_utils.py:1501]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0419 09:34:48.019101 139115631470400 maxtext_utils.py:1501]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   PartitionSpec('embed', 'vocab')
    Physical:  ('fsdp', None)

I0419 09:34:48.019155 139115631470400 maxtext_utils.py:1501]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   PartitionSpec('vocab', 'embed')
    Physical:  (None, 'fsdp')
I0419 09:34:48.517351 139115631470400 train.py:156] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0419 09:34:48.517452 139115631470400 train.py:156] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0419 09:34:48.532438 139115631470400 train.py:163] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0419 09:34:48.532500 139115631470400 train.py:163] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0419 09:34:59.311424 139115631470400 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.
I0419 09:34:59.312294 139115631470400 metric_logger.py:289] number parameters: 1.104 billion
I0419 09:35:10.288925 139115631470400 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0419 09:35:10.613421 139115631470400 checkpointing.py:776] Waited 0.32448863983154297 seconds for step 0 to finish before starting checkpointing.
I0419 09:35:10.615930 139115631470400 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0419 09:35:10.617675 139115631470400 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 0
I0419 09:35:10.620407 139115631470400 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/0.
I0419 09:35:11.615103 139115631470400 signaling_client.py:364] Using JaxDistributedSignalingClient
I0419 09:35:11.616133 139115631470400 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array.
I0419 09:35:11.616191 139115631470400 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
I0419 09:35:11.867300 139115631470400 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.252280s
I0419 09:35:11.867458 139115631470400 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.756 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5596847534179688 s) (per-host)
I0419 09:35:11.867507 139115631470400 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.559743s (batch_requests_ready=0.303880s, total_serialization_initiated=0.255797s, others=0.000066s)
I0419 09:35:11.867587 139115631470400 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.563849s (all_items=0.000021s, per_item={'items': '0.00002146'}, temp_paths=0.563827)
I0419 09:35:11.868507 138992792106752 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started.
I0419 09:35:11.868677 139115631470400 async_checkpointer.py:561] Finished blocking save. Time taken: 1.250936s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/0.
I0419 09:35:12.269924 139115631470400 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0419 09:35:12.270329 138988379674368 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save.
I0419 09:35:12.270495 139115631470400 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776591310.6159058, 'wait_for_prev_duration_secs': 7.271766662597656e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776591310.6177135, 'checkpointer_blocking_duration_secs': 1.2510673999786377, 'get_old_steps_start_time': 1776591311.868802, 'get_old_steps_duration_secs': 2.4080276489257812e-05, 'checkpoint_manager_blocking_start_time': 1776591310.6139913, 'checkpoint_manager_blocking_duration_secs': 1.6564640998840332}
I0419 09:35:12.270601 139115631470400 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0419 09:35:12.270653 139115631470400 max_utils.py:750] 
Memstats: After params initialized:
I0419 09:35:12.270711 139115631470400 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0))
I0419 09:35:12.270749 139115631470400 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0))
I0419 09:35:12.270776 139115631470400 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0))
I0419 09:35:12.270798 139115631470400 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0))
I0419 09:35:12.537911 139115631470400 metric_logger.py:185] completed step: 0, seconds: 10.977, TFLOP/s/device: 1.238, Tokens/s/device: 186.580, total_weights: 65536, loss: 10.874
I0419 09:35:12.706432 139115631470400 metric_logger.py:185] completed step: 1, seconds: 2.248, TFLOP/s/device: 6.045, Tokens/s/device: 911.155, total_weights: 65536, loss: 10.874
I0419 09:35:13.123876 139115631470400 metric_logger.py:185] completed step: 2, seconds: 0.021, TFLOP/s/device: 639.936, Tokens/s/device: 96458.176, total_weights: 65536, loss: 10.263
I0419 09:35:13.271364 139115631470400 metric_logger.py:185] completed step: 3, seconds: 0.418, TFLOP/s/device: 32.533, Tokens/s/device: 4903.663, total_weights: 65536, loss: 9.734
I0419 09:35:13.567349 139115631470400 metric_logger.py:185] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.766, Tokens/s/device: 13379.762, total_weights: 65536, loss: 9.278
I0419 09:35:13.572965 139115631470400 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.236, Tokens/s/device: 13902.843, total_weights: 65536, loss: 8.892
I0419 09:35:32.040617 139115631470400 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.850, Tokens/s/device: 6911.051, total_weights: 65536, loss: 8.592
I0419 09:35:32.188126 139115631470400 metric_logger.py:185] completed step: 7, seconds: 18.321, TFLOP/s/device: 0.742, Tokens/s/device: 111.784, total_weights: 65536, loss: 8.384
I0419 09:35:32.335900 139115631470400 metric_logger.py:185] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.453, Tokens/s/device: 13483.264, total_weights: 65536, loss: 8.255
I0419 09:35:32.483033 139115631470400 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0419 09:35:32.483690 139115631470400 checkpointing.py:776] Waited 0.0006778240203857422 seconds for step 9 to finish before starting checkpointing.
I0419 09:35:32.485808 139115631470400 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0419 09:35:33.566566    2649 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0419 09:35:35.910861 138993278621440 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/0/items/array_metadatas/process_3
I0419 09:36:07.258886 138992792106752 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 28.232 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 55.951061725616455 s) (per-host)
I0419 09:36:07.259006 138992792106752 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 55.390387s.
I0419 09:36:15.677781 138992792106752 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 63.809145s.
I0419 09:36:15.678089 138988379674368 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save.
I0419 09:36:15.678209 138988379674368 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save.
I0419 09:36:15.678276 138988379674368 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0419 09:36:15.679789 138988379674368 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0419 09:36:15.679969 139115631470400 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0419 09:36:15.680121 139115631470400 checkpoint_manager.py:1458] Waiting for previous save to complete took 43.194313 seconds. If this number is high, consider checkpointing less frequently.
I0419 09:36:15.681871 139115631470400 checkpoint_manager.py:1518] [process=3] Saving checkpoint at step 9
I0419 09:36:15.685259 139115631470400 async_checkpointer.py:452] [process=3] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/9.
I0419 09:36:16.667292 139115631470400 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array.
I0419 09:36:16.667390 139115631470400 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
I0419 09:36:16.718878 139115631470400 base_pytree_checkpoint_handler.py:153] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.053566s
I0419 09:36:16.719056 139115631470400 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/blocking_gbytes_per_sec: 4.660 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.33101344108581543 s) (per-host)
I0419 09:36:16.719107 139115631470400 base_pytree_checkpoint_handler.py:737] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.331074s (batch_requests_ready=0.275589s, total_serialization_initiated=0.055415s, others=0.000069s)
I0419 09:36:16.719195 139115631470400 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.335176s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.335158)
I0419 09:36:16.720086 138989495379712 async_checkpointer.py:79] [process=3][thread=async_save] Background save thread started.
I0419 09:36:16.720280 139115631470400 async_checkpointer.py:561] Finished blocking save. Time taken: 1.038337s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/9.
I0419 09:36:16.722177 139115631470400 checkpoint_manager.py:1566] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0419 09:36:16.722481 138988379674368 async_checkpointer.py:265] [process=3][thread=save_finalize] Waiting for background save thread=async_save.
I0419 09:36:16.722645 139115631470400 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776591332.4857802, 'wait_for_prev_duration_secs': 43.194313049316406, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776591375.6819103, 'checkpointer_blocking_duration_secs': 1.0384817123413086, 'get_old_steps_start_time': 1776591376.7204106, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1776591332.48392, 'checkpoint_manager_blocking_duration_secs': 44.23868942260742}
I0419 09:36:16.722746 139115631470400 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0419 09:36:16.722788 139115631470400 checkpoint_manager.py:2024] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0419 09:36:22.067828 138993278621440 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_false/checkpoints/9/items/array_metadatas/process_3
I0419 09:36:59.720708 138989495379712 base_pytree_checkpoint_handler.py:129] [process=3] /jax/checkpoint/write/gbytes_per_sec: 36.453 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.332637310028076 s) (per-host)
I0419 09:36:59.720813 138989495379712 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.000599s.
I0419 09:37:08.071437 138989495379712 async_checkpointer.py:144] [process=3][thread=async_save] Background save thread done. Time taken: 51.351207s.
I0419 09:37:08.071725 138988379674368 async_checkpointer.py:273] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save.
I0419 09:37:08.071850 138988379674368 async_checkpointer.py:283] [process=3][thread=save_finalize] No errors found in background save thread=async_save.
I0419 09:37:08.071900 138988379674368 checkpoint_manager.py:2133] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0419 09:37:08.073424 138988379674368 checkpoint_manager.py:2142] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0419 09:37:08.073602 139115631470400 checkpoint_manager.py:2036] [process=3][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0419 09:37:08.073740 139115631470400 checkpoint_manager.py:2013] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0419 09:37:08.074434 139115631470400 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.145, Tokens/s/device: 13889.077, total_weights: 65536, loss: 8.179
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Sun Apr 19 09:37:19 UTC 2026
EXIT_CODE=0