MaxView

‹ 01_baseCase: 02_synthetic03_dropout ›

Metrics: Linen vs NNX  ·  test/pipeline-scan-nnx

MetricLinen  5b21b66fcNNX  5b21b66fcDiff (NNX − Linen)
Parameters1.104 billion
Final loss8.1790
TFLOP/s92.252
Tok/s13905.2
Avg s/step2.741
Memory %1.38
JAX0.9.20.9.2

Diff = NNX value − Linen value. Green = NNX improved. Red = NNX regressed.

Linen  ·  5b21b66fc  ·  test_pipeline_scan_nnx_20260423_211617  ·  full log
XPK Start: Thu Apr 23 21:19:59 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. 
2026-04-23 21:20:24.017520: 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)
I0423 21:20:24.228363 133607239825216 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 21:20:33.269732 133607239825216 distributed.py:149] Starting JAX distributed service on [::]:8482
I0423 21:20:33.272058 133607239825216 distributed.py:172] Connecting to JAX distributed service on mt-02-synthetic-zqq4h-slice-job-0-0.mt-02-synthetic-zqq4h:8482
I0423 21:20:34.821344 133607239825216 max_utils.py:284] Jax distributed system initialized!
I0423 21:20:39.939511 133607239825216 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 21:20:39.939619 133607239825216 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 21:20:39.939661 133607239825216 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
I0423 21:20:39.939694 133607239825216 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 21:20:40.633132 133607239825216 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 21:20:40.633416 133607239825216 checkpointing.py:677] Setting up checkpoint logger...
I0423 21:20:40.633469 133607239825216 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0423 21:20:40.633514 133607239825216 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0423 21:20:40.633863 133607239825216 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 0x79834a91b5c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 21:20:43.580131 133607239825216 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0423 21:20:43.580376 133607239825216 checkpoint_manager.py:708] [process=4][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796ea46b7bc0>}, handler_registry=None
I0423 21:20:43.580617 133607239825216 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796ea46b7bc0>` 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`.
I0423 21:20:43.580665 133607239825216 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796ea46bacc0>` 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`.
I0423 21:20:43.580700 133607239825216 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 0x796ea46b7bc0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796ea46b7bc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796ea46bacc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796ea46bacc0>}).
I0423 21:20:43.581036 133607239825216 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0423 21:20:43.581107 133607239825216 async_checkpointer.py:192] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x796e8479f2e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0423 21:20:44.372901 133607239825216 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints
I0423 21:20:44.825129 133607239825216 checkpoint_manager.py:929] [process=4][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_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x796ea46b7f80>
I0423 21:20:44.825317 133607239825216 checkpointing.py:301] Checkpoint manager created!
I0423 21:20:45.741501 133607239825216 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 21:20:45.741617 133607239825216 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:20:46.121230 133607239825216 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 21:20:46.121324 133607239825216 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:20:46.137717 133607239825216 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 21:20:46.137775 133607239825216 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:20:46.161564 133607239825216 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 21:20:46.161639 133607239825216 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:20:46.178143 133607239825216 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 21:20:46.178203 133607239825216 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:20:46.194508 133607239825216 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0423 21:20:46.194566 133607239825216 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:20:46.219210 133607239825216 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 21:20:46.219283 133607239825216 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:20:46.240341 133607239825216 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 21:20:46.240407 133607239825216 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0423 21:20:46.451578 133607239825216 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 21:20:46.451684 133607239825216 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0423 21:20:47.876114 133607239825216 maxtext_utils.py:1668]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0423 21:20:47.876240 133607239825216 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)
I0423 21:20:47.876293 133607239825216 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)
I0423 21:20:47.876351 133607239825216 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')
I0423 21:20:47.876405 133607239825216 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)
I0423 21:20:47.876444 133607239825216 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)
I0423 21:20:47.876496 133607239825216 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)
I0423 21:20:47.876553 133607239825216 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')
I0423 21:20:47.876593 133607239825216 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)
I0423 21:20:47.876630 133607239825216 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)
I0423 21:20:47.876678 133607239825216 maxtext_utils.py:1668]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0423 21:20:47.876725 133607239825216 maxtext_utils.py:1668]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')

    Physical:  (None, 'fsdp')
I0423 21:20:48.373381 133607239825216 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0423 21:20:48.373477 133607239825216 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0423 21:20:48.388811 133607239825216 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0423 21:20:48.388879 133607239825216 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0423 21:20:59.440536 133607239825216 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.
I0423 21:20:59.441361 133607239825216 metric_logger.py:301] number parameters: 1.104 billion
I0423 21:21:11.092256 133607239825216 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0423 21:21:11.245368 133607239825216 checkpointing.py:776] Waited 0.15309476852416992 seconds for step 0 to finish before starting checkpointing.
I0423 21:21:11.247843 133607239825216 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 21:21:11.249505 133607239825216 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 0
I0423 21:21:11.250935 133607239825216 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/0.
I0423 21:21:11.588643 133607239825216 signaling_client.py:364] Using JaxDistributedSignalingClient
I0423 21:21:11.589541 133607239825216 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0423 21:21:11.589604 133607239825216 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
I0423 21:21:11.864377 133607239825216 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.275811s
I0423 21:21:11.864554 133607239825216 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 5.486 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28120946884155273 s) (per-host)
I0423 21:21:11.864605 133607239825216 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.281276s (batch_requests_ready=0.002190s, total_serialization_initiated=0.279012s, others=0.000074s)
I0423 21:21:11.864697 133607239825216 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.285382s (all_items=0.000028s, per_item={'items': '0.00002813'}, temp_paths=0.285353)
I0423 21:21:11.865543 133607239825216 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/0.
I0423 21:21:11.865822 133481406486272 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 21:41:11.865791
I0423 21:21:11.875376 133607239825216 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 21:21:11.875693 133479229220608 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0423 21:21:11.875853 133607239825216 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776979271.2478251, 'wait_for_prev_duration_secs': 7.295608520507812e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776979271.2495434, 'checkpointer_blocking_duration_secs': 0.6164374351501465, 'get_old_steps_start_time': 1776979271.8660073, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776979271.2458875, 'checkpoint_manager_blocking_duration_secs': 0.6299266815185547}
I0423 21:21:11.875967 133607239825216 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0423 21:21:11.876017 133607239825216 max_utils.py:750] 
Memstats: After params initialized:
I0423 21:21:11.876065 133607239825216 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0))
I0423 21:21:11.876096 133607239825216 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0))
I0423 21:21:11.876124 133607239825216 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0))
I0423 21:21:11.876150 133607239825216 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0))
I0423 21:21:12.186863 133607239825216 metric_logger.py:196] completed step: 0, seconds: 11.651, TFLOP/s/device: 1.166, Tokens/s/device: 175.782, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0423 21:21:12.362419 133607239825216 metric_logger.py:196] completed step: 1, seconds: 1.093, TFLOP/s/device: 12.430, Tokens/s/device: 1873.584, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0423 21:21:12.764382 133607239825216 metric_logger.py:196] completed step: 2, seconds: 0.029, TFLOP/s/device: 473.040, Tokens/s/device: 71301.744, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0423 21:21:12.911674 133607239825216 metric_logger.py:196] completed step: 3, seconds: 0.402, TFLOP/s/device: 33.821, Tokens/s/device: 5097.901, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0423 21:21:13.206593 133607239825216 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.561, Tokens/s/device: 13348.890, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0423 21:21:13.212363 133607239825216 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.557, Tokens/s/device: 13951.144, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0423 21:21:14.935742    2769 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0423 21:21:16.958244 133479237613312 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/0/items/array_metadatas/process_4
I0423 21:21:35.605430 133607239825216 metric_logger.py:196] completed step: 6, seconds: 0.295, TFLOP/s/device: 46.008, Tokens/s/device: 6934.874, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0423 21:21:35.752949 133607239825216 metric_logger.py:196] completed step: 7, seconds: 22.247, TFLOP/s/device: 0.611, Tokens/s/device: 92.059, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0423 21:21:35.900310 133607239825216 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.296, Tokens/s/device: 13459.605, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0423 21:21:36.046898 133607239825216 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0423 21:21:36.047578 133607239825216 checkpointing.py:776] Waited 0.0007171630859375 seconds for step 9 to finish before starting checkpointing.
I0423 21:21:36.049924 133607239825216 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 21:21:48.570528 133481406486272 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 42.707 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.98714089393616 s) (per-host)
I0423 21:21:48.570657 133481406486272 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.704711s.
I0423 21:21:56.794414 133481406486272 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 44.928452s.
I0423 21:21:56.794713 133479229220608 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 21:21:56.794838 133479229220608 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0423 21:21:56.794919 133479229220608 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0423 21:21:56.796918 133479229220608 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0423 21:21:56.797165 133607239825216 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0423 21:21:56.797309 133607239825216 checkpoint_manager.py:1452] Waiting for previous save to complete took 20.747385 seconds. If this number is high, consider checkpointing less frequently.
I0423 21:21:56.799219 133607239825216 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 9
I0423 21:21:56.801241 133607239825216 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/9.
I0423 21:21:57.494137 133607239825216 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0423 21:21:57.494235 133607239825216 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
I0423 21:21:57.539197 133607239825216 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046151s
I0423 21:21:57.539363 133607239825216 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 30.679 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05028104782104492 s) (per-host)
I0423 21:21:57.539413 133607239825216 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.050341s (batch_requests_ready=0.002459s, total_serialization_initiated=0.047817s, others=0.000065s)
I0423 21:21:57.539500 133607239825216 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054486s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.054469)
I0423 21:21:57.540259 133607239825216 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/9.
I0423 21:21:57.540568 133479229220608 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 21:41:57.540535
I0423 21:21:57.542560 133607239825216 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 21:21:57.542794 133479237613312 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0423 21:21:57.542909 133607239825216 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776979296.049884, 'wait_for_prev_duration_secs': 20.747384786605835, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776979316.7992575, 'checkpointer_blocking_duration_secs': 0.7414512634277344, 'get_old_steps_start_time': 1776979317.5407317, 'get_old_steps_duration_secs': 3.1948089599609375e-05, 'checkpoint_manager_blocking_start_time': 1776979296.0478125, 'checkpoint_manager_blocking_duration_secs': 21.49505376815796}
I0423 21:21:57.543054 133607239825216 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0423 21:21:57.543099 133607239825216 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 21:22:03.159888 133459158718208 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_02_synthetic/checkpoints/9/items/array_metadatas/process_4
I0423 21:22:39.036759 133479229220608 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 38.019 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.54763722419739 s) (per-host)
I0423 21:22:39.036905 133479229220608 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.496223s.
I0423 21:22:47.580919 133479229220608 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 50.040223s.
I0423 21:22:47.581133 133479237613312 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 21:22:47.581203 133479237613312 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0423 21:22:47.581256 133479237613312 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0423 21:22:47.583110 133479237613312 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0423 21:22:47.583291 133607239825216 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0423 21:22:47.583447 133607239825216 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 21:22:47.584373 133607239825216 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.252, Tokens/s/device: 13905.203, 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: Thu Apr 23 21:22:59 UTC 2026
EXIT_CODE=0
NNX  ·  5b21b66fc  ·  test_pipeline_scan_nnx_20260423_211617  ·  full log
XPK Start: Thu Apr 23 22:49:12 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. 
2026-04-23 22:49:37.614088: 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)
I0423 22:49:37.824217 140689412593472 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 22:49:46.866616 140689412593472 distributed.py:172] Connecting to JAX distributed service on mt-02-synthetic-1egrf-slice-job-0-0.mt-02-synthetic-1egrf:8482
I0423 22:49:47.033494 140689412593472 max_utils.py:284] Jax distributed system initialized!
I0423 22:49:52.297693 140689412593472 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 22:49:52.297799 140689412593472 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 22:49:52.297840 140689412593472 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
I0423 22:49:52.297874 140689412593472 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
Traceback (most recent call last):
  File "<frozen runpy>", line 198, in _run_module_as_main
  File "<frozen runpy>", line 88, in _run_code
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 744, in <module>
    app.run(main)
  File "/usr/local/lib/python3.12/site-packages/absl/app.py", line 367, in run
    _run_main(main, args)
  File "/usr/local/lib/python3.12/site-packages/absl/app.py", line 312, in _run_main
    sys.exit(main(argv))
             ^^^^^^^^^^
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 740, in main
    train_func()
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 730, in train_func
    run(config, recorder, diagnostic_config)
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 709, in run
    train_loop(config, recorder)
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 536, in train_loop
    ) = train_utils.setup_train_loop(config, recorder)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/deps/src/maxtext/utils/train_utils.py", line 218, in setup_train_loop
    raise NotImplementedError("Pure NNX support has not been implemented yet.")
NotImplementedError: Pure NNX support has not been implemented yet.
XPK End: Thu Apr 23 22:49:58 UTC 2026
EXIT_CODE=1