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/s92.185
Tok/s13895.1
Avg s/step2.801
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 21:33:49 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 21:34:14.532156: 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 21:34:14.741929 136086165927744 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 21:34:23.784517 136086165927744 distributed.py:149] Starting JAX distributed service on [::]:8482
I0423 21:34:23.786805 136086165927744 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-7498w-slice-job-0-0.mt-07-eval-7498w:8482
I0423 21:34:24.949290 136086165927744 max_utils.py:284] Jax distributed system initialized!
I0423 21:34:30.029500 136086165927744 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 21:34:30.029606 136086165927744 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 21:34:30.029647 136086165927744 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 21:34:30.029683 136086165927744 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 21:34:30.731509 136086165927744 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 21:34:30.731781 136086165927744 checkpointing.py:677] Setting up checkpoint logger...
I0423 21:34:30.731838 136086165927744 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0423 21:34:30.731883 136086165927744 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0423 21:34:30.732233 136086165927744 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 0x7bc44fc51ca0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 21:34:33.642268 136086165927744 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0423 21:34:33.642513 136086165927744 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 0x7bc3401797f0>}, handler_registry=None
I0423 21:34:33.642750 136086165927744 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bc3401797f0>` 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 21:34:33.642800 136086165927744 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb0ec3f3d10>` 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 21:34:33.642837 136086165927744 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 0x7bc3401797f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bc3401797f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb0ec3f3d10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb0ec3f3d10>}).
I0423 21:34:33.643168 136086165927744 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0423 21:34:33.643236 136086165927744 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7bb0547832e0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0423 21:34:34.720825 136086165927744 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_07_eval/checkpoints
I0423 21:34:34.748303 136086165927744 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_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bb0ec3f0e60>
I0423 21:34:34.748435 136086165927744 checkpointing.py:301] Checkpoint manager created!
I0423 21:34:35.683685 136086165927744 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 21:34:35.683824 136086165927744 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:34:36.069513 136086165927744 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 21:34:36.069610 136086165927744 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:34:36.086262 136086165927744 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 21:34:36.086323 136086165927744 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 21:34:36.110279 136086165927744 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 21:34:36.110354 136086165927744 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:34:36.127023 136086165927744 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 21:34:36.127085 136086165927744 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:34:36.143722 136086165927744 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 21:34:36.143784 136086165927744 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:34:36.169061 136086165927744 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 21:34:36.169152 136086165927744 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 21:34:36.190489 136086165927744 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 21:34:36.190553 136086165927744 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0423 21:34:36.403106 136086165927744 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 21:34:36.403216 136086165927744 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0423 21:34:37.843804 136086165927744 maxtext_utils.py:1668]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0423 21:34:37.843963 136086165927744 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 21:34:37.844023 136086165927744 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 21:34:37.844083 136086165927744 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 21:34:37.844159 136086165927744 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 21:34:37.844200 136086165927744 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 21:34:37.844273 136086165927744 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 21:34:37.844351 136086165927744 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 21:34:37.844422 136086165927744 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 21:34:37.844494 136086165927744 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 21:34:37.844578 136086165927744 maxtext_utils.py:1668]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0423 21:34:37.844654 136086165927744 maxtext_utils.py:1668]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0423 21:34:38.340607 136086165927744 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0423 21:34:38.340702 136086165927744 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0423 21:34:38.356384 136086165927744 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0423 21:34:38.356442 136086165927744 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0423 21:34:49.101897 136086165927744 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 21:34:49.102836 136086165927744 metric_logger.py:301] number parameters: 1.104 billion
I0423 21:35:00.309558 136086165927744 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0423 21:35:01.094040 136086165927744 checkpointing.py:776] Waited 0.7844595909118652 seconds for step 0 to finish before starting checkpointing.
I0423 21:35:01.096434 136086165927744 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 21:35:01.098027 136086165927744 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0423 21:35:01.099529 136086165927744 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/0.
I0423 21:35:01.430808 136086165927744 signaling_client.py:364] Using JaxDistributedSignalingClient
I0423 21:35:01.431860 136086165927744 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0423 21:35:01.431917 136086165927744 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 21:35:01.709522 136086165927744 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278789s
I0423 21:35:01.709706 136086165927744 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.426 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28431177139282227 s) (per-host)
I0423 21:35:01.709762 136086165927744 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284379s (batch_requests_ready=0.002334s, total_serialization_initiated=0.281969s, others=0.000076s)
I0423 21:35:01.709855 136086165927744 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288532s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.288514)
I0423 21:35:01.710651 136086165927744 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/0.
I0423 21:35:01.710988 135960588633856 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 21:55:01.710949
I0423 21:35:01.717650 136086165927744 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 21:35:01.717911 135957919557376 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 21:35:01.718065 136086165927744 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_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776980101.0964153, 'wait_for_prev_duration_secs': 6.008148193359375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776980101.0980642, 'checkpointer_blocking_duration_secs': 0.6130988597869873, 'get_old_steps_start_time': 1776980101.7111888, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776980101.0945597, 'checkpoint_manager_blocking_duration_secs': 0.6234681606292725}
I0423 21:35:01.718181 136086165927744 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0423 21:35:01.718232 136086165927744 max_utils.py:750] 
Memstats: After params initialized:
I0423 21:35:01.718278 136086165927744 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0423 21:35:01.718309 136086165927744 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0423 21:35:01.718334 136086165927744 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0423 21:35:01.718360 136086165927744 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0423 21:35:02.027401 136086165927744 metric_logger.py:196] completed step: 0, seconds: 11.207, TFLOP/s/device: 1.212, Tokens/s/device: 182.749, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0423 21:35:02.220364 136086165927744 metric_logger.py:196] completed step: 1, seconds: 1.716, TFLOP/s/device: 7.916, Tokens/s/device: 1193.258, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0423 21:35:02.637381 136086165927744 metric_logger.py:196] completed step: 2, seconds: 0.046, TFLOP/s/device: 296.876, Tokens/s/device: 44748.400, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0423 21:35:02.784795 136086165927744 metric_logger.py:196] completed step: 3, seconds: 0.418, TFLOP/s/device: 32.541, Tokens/s/device: 4904.884, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0423 21:35:03.079605 136086165927744 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.934, Tokens/s/device: 13405.159, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0423 21:35:03.085695 136086165927744 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.282, Tokens/s/device: 13909.736, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0423 21:35:05.459462    2780 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0423 21:35:07.476951 135957927950080 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_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/0/items/array_metadatas/process_6
I0423 21:35:25.364866 136086165927744 metric_logger.py:196] completed step: 6, seconds: 0.295, TFLOP/s/device: 45.981, Tokens/s/device: 6930.720, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0423 21:35:25.512365 136086165927744 metric_logger.py:196] completed step: 7, seconds: 22.133, TFLOP/s/device: 0.614, Tokens/s/device: 92.532, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0423 21:35:25.660045 136086165927744 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.458, Tokens/s/device: 13484.152, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0423 21:35:25.806453 136086165927744 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0423 21:35:25.807065 136086165927744 checkpointing.py:776] Waited 0.0006346702575683594 seconds for step 9 to finish before starting checkpointing.
I0423 21:35:25.809088 136086165927744 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 21:35:38.776212 135960588633856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 42.291 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.35077977180481 s) (per-host)
I0423 21:35:38.776342 135960588633856 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.065225s.
I0423 21:35:48.477355 135960588633856 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 46.766222s.
I0423 21:35:48.477661 135957919557376 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 21:35:48.477783 135957919557376 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 21:35:48.477837 135957919557376 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0423 21:35:48.480823 135957919557376 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0423 21:35:48.480953 136086165927744 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.
W0423 21:35:48.481029 136086165927744 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.671943 seconds. If this number is high, consider checkpointing less frequently.
I0423 21:35:48.483006 136086165927744 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9
I0423 21:35:48.484964 136086165927744 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/9.
I0423 21:35:48.767797 136086165927744 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0423 21:35:48.767888 136086165927744 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 21:35:48.807558 136086165927744 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040814s
I0423 21:35:48.807722 136086165927744 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.245 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04504537582397461 s) (per-host)
I0423 21:35:48.807772 136086165927744 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045105s (batch_requests_ready=0.002583s, total_serialization_initiated=0.042455s, others=0.000066s)
I0423 21:35:48.807856 136086165927744 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049216s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.049200)
I0423 21:35:48.808559 136086165927744 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.33 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/9.
I0423 21:35:48.808907 135957919557376 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 21:55:48.808871
I0423 21:35:48.810961 136086165927744 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 21:35:48.811222 135957927950080 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 21:35:48.811381 136086165927744 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_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776980125.8090587, 'wait_for_prev_duration_secs': 22.671943426132202, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776980148.4830449, 'checkpointer_blocking_duration_secs': 0.32601118087768555, 'get_old_steps_start_time': 1776980148.8090804, 'get_old_steps_duration_secs': 5.316734313964844e-05, 'checkpoint_manager_blocking_start_time': 1776980125.8073246, 'checkpoint_manager_blocking_duration_secs': 23.004022121429443}
I0423 21:35:48.811493 136086165927744 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0423 21:35:48.811539 136086165927744 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 21:35:54.001008 135932661446400 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_20260423_211617/linen_xpk_test_pipeline_scan_nnx_20260423_211617_07_eval/checkpoints/9/items/array_metadatas/process_6
I0423 21:36:29.924538 135957919557376 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.375 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.16182017326355 s) (per-host)
I0423 21:36:29.924669 135957919557376 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.115647s.
I0423 21:36:38.237619 135957919557376 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.428582s.
I0423 21:36:38.237895 135957927950080 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 21:36:38.238008 135957927950080 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 21:36:38.238057 135957927950080 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0423 21:36:38.240529 135957927950080 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0423 21:36:38.240714 136086165927744 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.
I0423 21:36:38.240879 136086165927744 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 21:36:38.241819 136086165927744 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.185, Tokens/s/device: 13895.108, 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 21:36:49 UTC 2026
EXIT_CODE=0
NNX  ·  5b21b66fc  ·  test_pipeline_scan_nnx_20260423_211617  ·  full log
XPK Start: Thu Apr 23 22:55:17 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 22:55:42.233131: 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 22:55:42.442698 132133738833728 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 22:55:51.485463 132133738833728 distributed.py:149] Starting JAX distributed service on [::]:8482
I0423 22:55:51.487756 132133738833728 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-dr47u-slice-job-0-0.mt-07-eval-dr47u:8482
I0423 22:55:52.502316 132133738833728 max_utils.py:284] Jax distributed system initialized!
I0423 22:55:58.736632 132133738833728 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 22:55:58.736738 132133738833728 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 22:55:58.736783 132133738833728 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 22:55:58.736829 132133738833728 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 22:56:07 UTC 2026
EXIT_CODE=1