MaxView

‹ —Case: 01_base02_synthetic ›

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

MetricLinen  5b21b66fcNNX  5b21b66fcDiff (NNX − Linen)
Parameters1.104 billion
Final loss8.1790
TFLOP/s92.170
Tok/s13892.9
Avg s/step2.859
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_20260422_212603  ·  full log
XPK Start: Wed Apr 22 21:26:47 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 21:27:12.463508: 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 21:27:12.672589 139280925579072 max_utils.py:273] Attempting to initialize the jax distributed system...
I0422 21:27:21.714119 139280925579072 distributed.py:149] Starting JAX distributed service on [::]:8482
I0422 21:27:21.716553 139280925579072 distributed.py:172] Connecting to JAX distributed service on mt-01-base-xyvl5-slice-job-0-0.mt-01-base-xyvl5:8482
I0422 21:27:25.109940 139280925579072 max_utils.py:284] Jax distributed system initialized!
I0422 21:27:31.314243 139280925579072 max_utils.py:800] System Information: Jax Version: 0.9.2
I0422 21:27:31.314351 139280925579072 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0422 21:27:31.314392 139280925579072 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 21:27:31.314429 139280925579072 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0422 21:27:32.008370 139280925579072 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0422 21:27:32.008646 139280925579072 checkpointing.py:677] Setting up checkpoint logger...
I0422 21:27:32.008714 139280925579072 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0422 21:27:32.008758 139280925579072 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0422 21:27:32.009096 139280925579072 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 0x7eac53d63440>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0422 21:27:36.098174 139280925579072 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0422 21:27:36.098406 139280925579072 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eab1aa0ed50>}, handler_registry=None
I0422 21:27:36.098641 139280925579072 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eab1aa0ed50>` 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 21:27:36.098706 139280925579072 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e98381e7740>` 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 21:27:36.098744 139280925579072 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 0x7eab1aa0ed50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eab1aa0ed50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e98381e7740>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e98381e7740>}).
I0422 21:27:36.099086 139280925579072 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0422 21:27:36.099158 139280925579072 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e980876f560> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0422 21:27:37.696708 139280925579072 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints
I0422 21:27:37.728235 139280925579072 checkpoint_manager.py:929] [process=5][thread=MainThread] CheckpointManager created,  primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e98381e3260>
I0422 21:27:37.728373 139280925579072 checkpointing.py:301] Checkpoint manager created!
I0422 21:27:38.645792 139280925579072 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0422 21:27:38.645905 139280925579072 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 21:27:39.026845 139280925579072 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 21:27:39.026946 139280925579072 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 21:27:39.043310 139280925579072 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0422 21:27:39.043373 139280925579072 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0422 21:27:39.067106 139280925579072 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 21:27:39.067175 139280925579072 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 21:27:39.083717 139280925579072 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 21:27:39.083775 139280925579072 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 21:27:39.100219 139280925579072 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 21:27:39.100280 139280925579072 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 21:27:39.125052 139280925579072 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0422 21:27:39.125126 139280925579072 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0422 21:27:39.146209 139280925579072 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0422 21:27:39.146274 139280925579072 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0422 21:27:39.355726 139280925579072 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0422 21:27:39.355832 139280925579072 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0422 21:27:40.787019 139280925579072 maxtext_utils.py:1668]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0422 21:27:40.787148 139280925579072 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 21:27:40.787202 139280925579072 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 21:27:40.787258 139280925579072 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 21:27:40.787318 139280925579072 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 21:27:40.787363 139280925579072 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 21:27:40.787418 139280925579072 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 21:27:40.787469 139280925579072 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 21:27:40.787508 139280925579072 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 21:27:40.787543 139280925579072 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 21:27:40.787590 139280925579072 maxtext_utils.py:1668]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0422 21:27:40.787636 139280925579072 maxtext_utils.py:1668]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0422 21:27:41.281563 139280925579072 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0422 21:27:41.281666 139280925579072 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0422 21:27:41.297120 139280925579072 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0422 21:27:41.297184 139280925579072 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0422 21:27:52.359405 139280925579072 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 21:27:52.360165 139280925579072 metric_logger.py:301] number parameters: 1.104 billion
I0422 21:28:04.055583 139280925579072 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0422 21:28:04.208370 139280925579072 checkpointing.py:776] Waited 0.1527721881866455 seconds for step 0 to finish before starting checkpointing.
I0422 21:28:04.210893 139280925579072 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 21:28:04.212787 139280925579072 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 0
I0422 21:28:04.214169 139280925579072 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints/0.
I0422 21:28:04.949516 139280925579072 signaling_client.py:364] Using JaxDistributedSignalingClient
I0422 21:28:04.950554 139280925579072 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0422 21:28:04.950613 139280925579072 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 21:28:05.224290 139280925579072 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274845s
I0422 21:28:05.224479 139280925579072 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 5.502 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2803635597229004 s) (per-host)
I0422 21:28:05.224542 139280925579072 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280442s (batch_requests_ready=0.002225s, total_serialization_initiated=0.278129s, others=0.000087s)
I0422 21:28:05.224642 139280925579072 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.284546s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.284528)
I0422 21:28:05.225456 139280925579072 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints/0.
I0422 21:28:05.225729 139151814936320 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 21:48:05.225704
I0422 21:28:05.659262 139280925579072 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 21:28:05.659700 139150178023168 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save.
I0422 21:28:05.659898 139280925579072 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776893284.2108738, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776893284.2128272, 'checkpointer_blocking_duration_secs': 1.0130133628845215, 'get_old_steps_start_time': 1776893285.2258642, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776893284.208917, 'checkpoint_manager_blocking_duration_secs': 1.4509377479553223}
I0422 21:28:05.660009 139280925579072 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0422 21:28:05.660062 139280925579072 max_utils.py:750] 
Memstats: After params initialized:
I0422 21:28:05.660119 139280925579072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0))
I0422 21:28:05.660154 139280925579072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0))
I0422 21:28:05.660182 139280925579072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0))
I0422 21:28:05.660209 139280925579072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0))
I0422 21:28:05.971621 139280925579072 metric_logger.py:196] completed step: 0, seconds: 11.695, TFLOP/s/device: 1.162, Tokens/s/device: 175.113, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0422 21:28:06.145855 139280925579072 metric_logger.py:196] completed step: 1, seconds: 1.914, TFLOP/s/device: 7.097, Tokens/s/device: 1069.793, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0422 21:28:06.989636 139280925579072 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 498.629, Tokens/s/device: 75158.721, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0422 21:28:07.137167 139280925579072 metric_logger.py:196] completed step: 3, seconds: 0.844, TFLOP/s/device: 16.102, Tokens/s/device: 2427.138, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0422 21:28:07.431968 139280925579072 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.743, Tokens/s/device: 13376.266, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0422 21:28:07.437826 139280925579072 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.236, Tokens/s/device: 13902.843, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0422 21:28:08.817476    2821 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0422 21:28:11.173067 139150186415872 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints/0/items/array_metadatas/process_5
I0422 21:28:29.634977 139280925579072 metric_logger.py:196] completed step: 6, seconds: 0.295, TFLOP/s/device: 45.998, Tokens/s/device: 6933.371, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0422 21:28:29.782476 139280925579072 metric_logger.py:196] completed step: 7, seconds: 22.050, TFLOP/s/device: 0.616, Tokens/s/device: 92.878, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0422 21:28:29.929643 139280925579072 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.284, Tokens/s/device: 13457.836, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0422 21:28:30.076454 139280925579072 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0422 21:28:30.077120 139280925579072 checkpointing.py:776] Waited 0.0006885528564453125 seconds for step 9 to finish before starting checkpointing.
I0422 21:28:30.079236 139280925579072 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 21:28:42.039739 139151814936320 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 42.582 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.095590591430664 s) (per-host)
I0422 21:28:42.039867 139151814936320 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.814059s.
I0422 21:28:51.438332 139151814936320 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 46.212508s.
I0422 21:28:51.438632 139150178023168 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 21:28:51.438791 139150178023168 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save.
I0422 21:28:51.438854 139150178023168 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0422 21:28:51.440372 139150178023168 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0422 21:28:51.440550 139280925579072 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0422 21:28:51.440695 139280925579072 checkpoint_manager.py:1452] Waiting for previous save to complete took 21.361459 seconds. If this number is high, consider checkpointing less frequently.
I0422 21:28:51.444931 139280925579072 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 9
I0422 21:28:51.446866 139280925579072 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints/9.
I0422 21:28:52.157315 139280925579072 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0422 21:28:52.157407 139280925579072 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 21:28:52.193891 139280925579072 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037687s
I0422 21:28:52.194064 139280925579072 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 37.467 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04117155075073242 s) (per-host)
I0422 21:28:52.194116 139280925579072 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.041234s (batch_requests_ready=0.001799s, total_serialization_initiated=0.039364s, others=0.000071s)
I0422 21:28:52.194210 139280925579072 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.045223s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.045208)
I0422 21:28:52.195011 139280925579072 event_tracking.py:125] [process=5] [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_01_base/checkpoints/9.
I0422 21:28:52.195349 139150178023168 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 21:48:52.195312
I0422 21:28:52.198851 139280925579072 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0422 21:28:52.199113 139150186415872 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save.
I0422 21:28:52.199275 139280925579072 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776893310.079206, 'wait_for_prev_duration_secs': 21.36145853996277, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776893331.4449708, 'checkpointer_blocking_duration_secs': 0.7505180835723877, 'get_old_steps_start_time': 1776893332.195513, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776893310.0773528, 'checkpoint_manager_blocking_duration_secs': 22.121887922286987}
I0422 21:28:52.199383 139280925579072 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0422 21:28:52.199429 139280925579072 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0422 21:28:59.634391 139128888882944 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260422_212603/linen_xpk_test_pipeline_scan_nnx_20260422_212603_01_base/checkpoints/9/items/array_metadatas/process_5
I0422 21:29:34.547323 139150178023168 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.260 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.39439129829407 s) (per-host)
I0422 21:29:34.547448 139150178023168 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.351987s.
I0422 21:29:41.665764 139150178023168 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 49.470289s.
I0422 21:29:41.666078 139150186415872 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save.
I0422 21:29:41.666197 139150186415872 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save.
I0422 21:29:41.666246 139150186415872 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0422 21:29:41.668229 139150186415872 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0422 21:29:41.668412 139280925579072 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0422 21:29:41.668565 139280925579072 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0422 21:29:41.669425 139280925579072 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.170, Tokens/s/device: 13892.940, 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: Wed Apr 22 21:29:52 UTC 2026
EXIT_CODE=0
NNX  ·  5b21b66fc  ·  test_pipeline_scan_nnx_20260422_212603  ·  full log
XPK Start: Wed Apr 22 23:03:34 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 23:03:59.343150: 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 23:03:59.556535 132459360966464 max_utils.py:273] Attempting to initialize the jax distributed system...
I0422 23:04:08.597315 132459360966464 distributed.py:149] Starting JAX distributed service on [::]:8482
I0422 23:04:08.599550 132459360966464 distributed.py:172] Connecting to JAX distributed service on mt-01-base-doah7-slice-job-0-0.mt-01-base-doah7:8482
I0422 23:04:09.888985 132459360966464 max_utils.py:284] Jax distributed system initialized!
I0422 23:04:16.083166 132459360966464 max_utils.py:800] System Information: Jax Version: 0.9.2
I0422 23:04:16.083275 132459360966464 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0422 23:04:16.083317 132459360966464 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 23:04:16.083354 132459360966464 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: Wed Apr 22 23:04:24 UTC 2026
EXIT_CODE=1