MaxView

‹ 06_grad_accumCase: 07_eval08_checkpoint_async_true_resume ›

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

MetricLinen  5b21b66fcNNX  5b21b66fcDiff (NNX − Linen)
Parameters1.104 billion
Final loss8.1790
TFLOP/s92.192
Tok/s13896.2
Avg s/step2.712
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_20260424_200834  ·  full log
XPK Start: Fri Apr 24 20:26:07 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. 
2026-04-24 20:26:32.115488: 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)
I0424 20:26:32.326112 137430717458240 max_utils.py:273] Attempting to initialize the jax distributed system...
I0424 20:26:41.368393 137430717458240 distributed.py:149] Starting JAX distributed service on [::]:8482
I0424 20:26:41.370835 137430717458240 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-h142m-slice-job-0-0.mt-07-eval-h142m:8482
I0424 20:26:43.002118 137430717458240 max_utils.py:284] Jax distributed system initialized!
I0424 20:26:48.151125 137430717458240 max_utils.py:800] System Information: Jax Version: 0.9.2
I0424 20:26:48.151229 137430717458240 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0424 20:26:48.151272 137430717458240 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
I0424 20:26:48.151306 137430717458240 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0424 20:26:48.842654 137430717458240 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0424 20:26:48.842934 137430717458240 checkpointing.py:677] Setting up checkpoint logger...
I0424 20:26:48.842989 137430717458240 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0424 20:26:48.843031 137430717458240 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0424 20:26:48.843411 137430717458240 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 0x7cfd5d748f50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0424 20:26:51.668978 137430717458240 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0424 20:26:51.669239 137430717458240 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ce9c42cb200>}, handler_registry=None
I0424 20:26:51.669476 137430717458240 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ce9c42cb200>` 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`.
I0424 20:26:51.669530 137430717458240 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ce9c42ceb10>` 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`.
I0424 20:26:51.669567 137430717458240 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 0x7ce9c42cb200>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ce9c42cb200>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ce9c42ceb10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ce9c42ceb10>}).
I0424 20:26:51.669885 137430717458240 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0424 20:26:51.669955 137430717458240 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ce8dc7e72e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0424 20:26:52.856317 137430717458240 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints
I0424 20:26:52.982724 137430717458240 checkpoint_manager.py:929] [process=6][thread=MainThread] CheckpointManager created,  primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ce9c4234530>
I0424 20:26:52.982899 137430717458240 checkpointing.py:301] Checkpoint manager created!
I0424 20:26:53.923280 137430717458240 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0424 20:26:53.923399 137430717458240 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 20:26:54.307383 137430717458240 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0424 20:26:54.307479 137430717458240 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 20:26:54.324139 137430717458240 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0424 20:26:54.324198 137430717458240 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0424 20:26:54.348222 137430717458240 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 20:26:54.348291 137430717458240 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 20:26:54.365016 137430717458240 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 20:26:54.365078 137430717458240 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 20:26:54.381755 137430717458240 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0424 20:26:54.381817 137430717458240 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 20:26:54.407032 137430717458240 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0424 20:26:54.407115 137430717458240 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0424 20:26:54.428502 137430717458240 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0424 20:26:54.428572 137430717458240 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0424 20:26:54.647597 137430717458240 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0424 20:26:54.647712 137430717458240 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0424 20:26:56.098194 137430717458240 maxtext_utils.py:1668]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0424 20:26:56.098325 137430717458240 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)
I0424 20:26:56.098379 137430717458240 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)
I0424 20:26:56.098438 137430717458240 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')
I0424 20:26:56.098491 137430717458240 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)
I0424 20:26:56.098529 137430717458240 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)
I0424 20:26:56.098582 137430717458240 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)
I0424 20:26:56.098631 137430717458240 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')
I0424 20:26:56.098672 137430717458240 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)
I0424 20:26:56.098709 137430717458240 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)
I0424 20:26:56.098757 137430717458240 maxtext_utils.py:1668]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0424 20:26:56.098804 137430717458240 maxtext_utils.py:1668]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0424 20:26:56.596509 137430717458240 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0424 20:26:56.596607 137430717458240 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0424 20:26:56.612497 137430717458240 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0424 20:26:56.612560 137430717458240 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0424 20:27:07.723808 137430717458240 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.
I0424 20:27:07.724683 137430717458240 metric_logger.py:301] number parameters: 1.104 billion
I0424 20:27:19.262948 137430717458240 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0424 20:27:19.416918 137430717458240 checkpointing.py:776] Waited 0.15395140647888184 seconds for step 0 to finish before starting checkpointing.
I0424 20:27:19.419230 137430717458240 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0424 20:27:19.420762 137430717458240 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0424 20:27:19.422209 137430717458240 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/0.
I0424 20:27:20.154137 137430717458240 signaling_client.py:364] Using JaxDistributedSignalingClient
I0424 20:27:20.155084 137430717458240 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0424 20:27:20.155156 137430717458240 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
I0424 20:27:20.435945 137430717458240 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.281894s
I0424 20:27:20.436133 137430717458240 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.367 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2874166965484619 s) (per-host)
I0424 20:27:20.436188 137430717458240 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287482s (batch_requests_ready=0.002325s, total_serialization_initiated=0.285068s, others=0.000088s)
I0424 20:27:20.436285 137430717458240 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291837s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.291818)
I0424 20:27:20.437155 137430717458240 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/0.
I0424 20:27:20.437445 137304991954688 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 20:47:20.437412
I0424 20:27:20.439396 137430717458240 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0424 20:27:20.439682 137301744420608 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0424 20:27:20.439807 137430717458240 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777062439.419211, 'wait_for_prev_duration_secs': 6.127357482910156e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777062439.4208007, 'checkpointer_blocking_duration_secs': 1.016800880432129, 'get_old_steps_start_time': 1777062440.4376285, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1777062439.4174564, 'checkpoint_manager_blocking_duration_secs': 1.0223166942596436}
I0424 20:27:20.439953 137430717458240 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0424 20:27:20.440005 137430717458240 max_utils.py:750] 
Memstats: After params initialized:
I0424 20:27:20.440052 137430717458240 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0424 20:27:20.440084 137430717458240 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0424 20:27:20.440131 137430717458240 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0424 20:27:20.440160 137430717458240 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0424 20:27:20.754989 137430717458240 metric_logger.py:196] completed step: 0, seconds: 11.538, TFLOP/s/device: 1.178, Tokens/s/device: 177.498, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0424 20:27:20.930838 137430717458240 metric_logger.py:196] completed step: 1, seconds: 1.491, TFLOP/s/device: 9.115, Tokens/s/device: 1373.952, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0424 20:27:21.357304 137430717458240 metric_logger.py:196] completed step: 2, seconds: 0.027, TFLOP/s/device: 499.031, Tokens/s/device: 75219.451, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0424 20:27:21.504276 137430717458240 metric_logger.py:196] completed step: 3, seconds: 0.428, TFLOP/s/device: 31.773, Tokens/s/device: 4789.120, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0424 20:27:21.799602 137430717458240 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.660, Tokens/s/device: 13363.872, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0424 20:27:21.805160 137430717458240 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.419, Tokens/s/device: 13930.457, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0424 20:27:28.803683    2763 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0424 20:27:30.847399 137302290458368 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/0/items/array_metadatas/process_6
I0424 20:27:43.517954 137430717458240 metric_logger.py:196] completed step: 6, seconds: 0.295, TFLOP/s/device: 45.980, Tokens/s/device: 6930.626, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0424 20:27:43.665367 137430717458240 metric_logger.py:196] completed step: 7, seconds: 21.566, TFLOP/s/device: 0.630, Tokens/s/device: 94.963, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0424 20:27:43.812707 137430717458240 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.616, Tokens/s/device: 13507.898, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0424 20:27:43.959268 137430717458240 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0424 20:27:43.959934 137430717458240 checkpointing.py:776] Waited 0.0006883144378662109 seconds for step 9 to finish before starting checkpointing.
I0424 20:27:43.961966 137430717458240 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0424 20:28:02.068587 137304991954688 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.681 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.919820070266724 s) (per-host)
I0424 20:28:02.068716 137304991954688 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.631160s.
I0424 20:28:10.140753 137304991954688 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.703181s.
I0424 20:28:10.141030 137301744420608 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0424 20:28:10.141182 137301744420608 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0424 20:28:10.141241 137301744420608 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0424 20:28:10.143160 137301744420608 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0424 20:28:10.143362 137430717458240 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0424 20:28:10.143502 137430717458240 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.181536 seconds. If this number is high, consider checkpointing less frequently.
I0424 20:28:10.145377 137430717458240 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9
I0424 20:28:10.147447 137430717458240 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/9.
I0424 20:28:10.863011 137430717458240 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0424 20:28:10.863118 137430717458240 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
I0424 20:28:10.907299 137430717458240 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045459s
I0424 20:28:10.907473 137430717458240 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.040 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04969596862792969 s) (per-host)
I0424 20:28:10.907524 137430717458240 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049757s (batch_requests_ready=0.002597s, total_serialization_initiated=0.047091s, others=0.000069s)
I0424 20:28:10.907618 137430717458240 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053985s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.053969)
I0424 20:28:10.908351 137430717458240 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/9.
I0424 20:28:10.908701 137301744420608 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 20:48:10.908663
I0424 20:28:10.910642 137430717458240 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0424 20:28:10.910922 137277237860096 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0424 20:28:10.911091 137430717458240 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777062463.9619358, 'wait_for_prev_duration_secs': 26.181535720825195, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777062490.145415, 'checkpointer_blocking_duration_secs': 0.7634334564208984, 'get_old_steps_start_time': 1777062490.9088721, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1777062463.960177, 'checkpoint_manager_blocking_duration_secs': 26.950881004333496}
I0424 20:28:10.911209 137430717458240 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0424 20:28:10.911259 137430717458240 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0424 20:28:15.970888 137302290458368 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260424_200834/linen_xpk_test_pipeline_scan_nnx_20260424_200834_07_eval/checkpoints/9/items/array_metadatas/process_6
I0424 20:28:52.611607 137301744420608 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.831 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.753790855407715 s) (per-host)
I0424 20:28:52.611735 137301744420608 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.702921s.
I0424 20:29:01.324017 137301744420608 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.415187s.
I0424 20:29:01.324344 137277237860096 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0424 20:29:01.324468 137277237860096 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0424 20:29:01.324518 137277237860096 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0424 20:29:01.325920 137277237860096 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0424 20:29:01.326120 137430717458240 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0424 20:29:01.326280 137430717458240 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0424 20:29:01.327190 137430717458240 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.192, Tokens/s/device: 13896.240, 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: Fri Apr 24 20:29:12 UTC 2026
EXIT_CODE=0
NNX  ·  5b21b66fc  ·  test_pipeline_scan_nnx_20260424_200834  ·  full log
XPK Start: Fri Apr 24 21:46:47 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. 
2026-04-24 21:47:12.754395: 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)
I0424 21:47:12.968301 138664910477120 max_utils.py:273] Attempting to initialize the jax distributed system...
I0424 21:47:22.008414 138664910477120 distributed.py:149] Starting JAX distributed service on [::]:8482
I0424 21:47:22.010883 138664910477120 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-6bpfr-slice-job-0-0.mt-07-eval-6bpfr:8482
I0424 21:47:22.975349 138664910477120 max_utils.py:284] Jax distributed system initialized!
I0424 21:47:29.078036 138664910477120 max_utils.py:800] System Information: Jax Version: 0.9.2
I0424 21:47:29.078166 138664910477120 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0424 21:47:29.078208 138664910477120 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
I0424 21:47:29.078245 138664910477120 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: Fri Apr 24 21:47:37 UTC 2026
EXIT_CODE=1