MaxView

‹ 16_zarr3_false_resumeCase: 16_zarr3_false_save16_zarr3_true_resume ›

Metrics: Linen vs NNX  ·  main

MetricLinen  574ad3fb9NNX  574ad3fb9Diff (NNX − Linen)
Parameters1.104 billion
Final loss7.1810
TFLOP/s92.056
Tok/s13875.6
Avg s/step8.330
Memory %1.38
JAX0.9.20.9.2

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

Linen  ·  574ad3fb9  ·  main_20260420_060028  ·  full log
XPK Start: Mon Apr 20 07:26:09 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-20 07:26:33.179830: 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)
I0420 07:26:33.713930 134540780734272 max_utils.py:273] Attempting to initialize the jax distributed system...
I0420 07:26:42.755499 134540780734272 distributed.py:149] Starting JAX distributed service on [::]:8482
I0420 07:26:42.757705 134540780734272 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-sv-umb3t-slice-job-0-0.mt-16-zarr3-false-sv-umb3t:8482
I0420 07:26:43.914508 134540780734272 max_utils.py:284] Jax distributed system initialized!
I0420 07:26:50.227434 134540780734272 max_utils.py:800] System Information: Jax Version: 0.9.2
I0420 07:26:50.227535 134540780734272 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0420 07:26:50.227577 134540780734272 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
I0420 07:26:50.227611 134540780734272 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0420 07:26:50.916675 134540780734272 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0420 07:26:50.916951 134540780734272 checkpointing.py:677] Setting up checkpoint logger...
I0420 07:26:50.917004 134540780734272 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False
I0420 07:26:50.917049 134540780734272 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0420 07:26:50.917410 134540780734272 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7a5c949190a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0420 07:26:53.906528 134540780734272 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0420 07:26:53.906868 134540780734272 checkpoint_manager.py:708] [process=4][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a476c428140>}, handler_registry=None
I0420 07:26:53.907101 134540780734272 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a476c428140>` 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`.
I0420 07:26:53.907150 134540780734272 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a48941190a0>` 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`.
I0420 07:26:53.907185 134540780734272 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 0x7a476c428140>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a476c428140>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a48941190a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a48941190a0>}).
I0420 07:26:53.907524 134540780734272 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0420 07:26:53.907593 134540780734272 async_checkpointer.py:192] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a485c463880> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0420 07:26:55.045131 134540780734272 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints
I0420 07:26:55.047463 134540780734272 checkpoint_manager.py:929] [process=4][thread=MainThread] CheckpointManager created,  primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a4894114f20>
I0420 07:26:55.047572 134540780734272 checkpointing.py:301] Checkpoint manager created!
I0420 07:26:55.962185 134540780734272 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0420 07:26:55.962301 134540780734272 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 07:26:56.340510 134540780734272 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 07:26:56.340600 134540780734272 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 07:26:56.356701 134540780734272 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0420 07:26:56.356756 134540780734272 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0420 07:26:56.381006 134540780734272 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 07:26:56.381075 134540780734272 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 07:26:56.397522 134540780734272 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 07:26:56.397586 134540780734272 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 07:26:56.413791 134540780734272 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0420 07:26:56.413855 134540780734272 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 07:26:56.438545 134540780734272 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0420 07:26:56.438617 134540780734272 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0420 07:26:56.459491 134540780734272 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0420 07:26:56.459557 134540780734272 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0420 07:26:56.673900 134540780734272 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0420 07:26:56.674006 134540780734272 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0420 07:26:58.098365 134540780734272 maxtext_utils.py:1654]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0420 07:26:58.098493 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0420 07:26:58.098545 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0420 07:26:58.098602 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   P('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0420 07:26:58.098651 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0420 07:26:58.098688 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0420 07:26:58.098739 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0420 07:26:58.098794 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   P('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0420 07:26:58.098836 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0420 07:26:58.098873 134540780734272 maxtext_utils.py:1654]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0420 07:26:58.098920 134540780734272 maxtext_utils.py:1654]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0420 07:26:58.098968 134540780734272 maxtext_utils.py:1654]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0420 07:26:58.586974 134540780734272 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0420 07:26:58.587069 134540780734272 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0420 07:26:58.602346 134540780734272 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0420 07:26:58.602404 134540780734272 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0420 07:27:09.519072 134540780734272 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.
I0420 07:27:09.519877 134540780734272 metric_logger.py:301] number parameters: 1.104 billion
I0420 07:27:20.698709 134540780734272 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0420 07:27:20.902393 134540780734272 checkpointing.py:776] Waited 0.20364069938659668 seconds for step 0 to finish before starting checkpointing.
I0420 07:27:20.904982 134540780734272 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 07:27:20.906829 134540780734272 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 0
I0420 07:27:20.908301 134540780734272 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/0.
I0420 07:27:21.632758 134540780734272 signaling_client.py:364] Using JaxDistributedSignalingClient
I0420 07:27:21.633712 134540780734272 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0420 07:27:21.633770 134540780734272 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
I0420 07:27:21.914844 134540780734272 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282157s
I0420 07:27:21.915016 134540780734272 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 5.361 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28773975372314453 s) (per-host)
I0420 07:27:21.915069 134540780734272 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287802s (batch_requests_ready=0.002345s, total_serialization_initiated=0.285386s, others=0.000071s)
I0420 07:27:21.915160 134540780734272 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.291993s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.291974)
I0420 07:27:21.915951 134540780734272 event_tracking.py:125] [process=4] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/0.
I0420 07:27:21.916286 134412607928064 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 07:47:21.916248
I0420 07:27:21.918202 134540780734272 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 07:27:21.918496 134410976347904 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0420 07:27:21.918644 134540780734272 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776670040.9049647, 'wait_for_prev_duration_secs': 6.031990051269531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776670040.906868, 'checkpointer_blocking_duration_secs': 1.0095796585083008, 'get_old_steps_start_time': 1776670041.9164724, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776670040.9029756, 'checkpoint_manager_blocking_duration_secs': 1.0156335830688477}
I0420 07:27:21.918744 134540780734272 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0420 07:27:21.918794 134540780734272 max_utils.py:750] 
Memstats: After params initialized:
I0420 07:27:21.918838 134540780734272 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0))
I0420 07:27:21.918868 134540780734272 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0))
I0420 07:27:21.918893 134540780734272 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0))
I0420 07:27:21.918915 134540780734272 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0))
I0420 07:27:22.233490 134540780734272 metric_logger.py:196] completed step: 0, seconds: 11.179, TFLOP/s/device: 1.215, Tokens/s/device: 183.205, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0420 07:27:22.403717 134540780734272 metric_logger.py:196] completed step: 1, seconds: 1.533, TFLOP/s/device: 8.862, Tokens/s/device: 1335.735, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0420 07:27:22.809217 134540780734272 metric_logger.py:196] completed step: 2, seconds: 0.023, TFLOP/s/device: 591.903, Tokens/s/device: 89218.035, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719
I0420 07:27:22.956735 134540780734272 metric_logger.py:196] completed step: 3, seconds: 0.406, TFLOP/s/device: 33.483, Tokens/s/device: 5046.933, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506
I0420 07:27:22.961368 134540780734272 checkpointing.py:772] Waiting for step 5 to finish before checkpoint...
I0420 07:27:23.251269 134540780734272 checkpointing.py:776] Waited 0.28986477851867676 seconds for step 5 to finish before starting checkpointing.
I0420 07:27:23.254173 134540780734272 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 07:27:25.112023    2793 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0420 07:27:27.534164 134410997036800 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/0/items/array_metadatas/process_4
I0420 07:28:03.479656 134412607928064 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.742 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.85233950614929 s) (per-host)
I0420 07:28:03.479776 134412607928064 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.563368s.
I0420 07:28:12.719443 134412607928064 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 50.803017s.
I0420 07:28:12.719766 134410976347904 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 07:28:12.719883 134410976347904 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0420 07:28:12.719935 134410976347904 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0420 07:28:12.721817 134410976347904 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0420 07:28:12.721999 134540780734272 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0420 07:28:12.722138 134540780734272 checkpoint_manager.py:1452] Waiting for previous save to complete took 49.467965 seconds. If this number is high, consider checkpointing less frequently.
I0420 07:28:12.723950 134540780734272 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 5
I0420 07:28:12.726321 134540780734272 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/5.
I0420 07:28:13.042485 134540780734272 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0420 07:28:13.042581 134540780734272 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
I0420 07:28:13.076297 134540780734272 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034917s
I0420 07:28:13.076475 134540780734272 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 40.121 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03844785690307617 s) (per-host)
I0420 07:28:13.076527 134540780734272 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038510s (batch_requests_ready=0.001881s, total_serialization_initiated=0.036560s, others=0.000069s)
I0420 07:28:13.076628 134540780734272 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043552s (all_items=0.000015s, per_item={'items': '0.00001550'}, temp_paths=0.043536)
I0420 07:28:13.077241 134540780734272 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/5.
I0420 07:28:13.077511 134412588517120 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 07:48:13.077484
I0420 07:28:13.086063 134540780734272 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 07:28:13.086267 134410976347904 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0420 07:28:13.086391 134540780734272 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776670043.2541447, 'wait_for_prev_duration_secs': 49.46796536445618, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776670092.72399, 'checkpointer_blocking_duration_secs': 0.3536231517791748, 'get_old_steps_start_time': 1776670093.0776339, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776670043.2519526, 'checkpoint_manager_blocking_duration_secs': 49.83440613746643}
I0420 07:28:13.086499 134540780734272 checkpointing.py:408] Started an asynchronous checkpoint save for step 5
I0420 07:28:13.087275 134540780734272 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.766, Tokens/s/device: 13379.762, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355
I0420 07:28:13.093115 134540780734272 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.127, Tokens/s/device: 13886.440, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467
I0420 07:28:18.691542 134405719906048 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/5/items/array_metadatas/process_4
I0420 07:28:35.520355 134540780734272 metric_logger.py:196] completed step: 6, seconds: 50.130, TFLOP/s/device: 0.271, Tokens/s/device: 40.854, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740
I0420 07:28:35.667832 134540780734272 metric_logger.py:196] completed step: 7, seconds: 22.281, TFLOP/s/device: 0.610, Tokens/s/device: 91.917, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151
I0420 07:28:35.815321 134540780734272 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.180, Tokens/s/device: 13442.113, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098
I0420 07:28:35.962010 134540780734272 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0420 07:28:35.962703 134540780734272 checkpointing.py:776] Waited 0.0007138252258300781 seconds for step 9 to finish before starting checkpointing.
I0420 07:28:35.964803 134540780734272 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 07:28:55.811905 134412588517120 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 36.929 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.77383875846863 s) (per-host)
I0420 07:28:55.812028 134412588517120 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.734439s.
I0420 07:29:04.008492 134412588517120 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 50.930889s.
I0420 07:29:04.008739 134410976347904 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 07:29:04.008829 134410976347904 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0420 07:29:04.008876 134410976347904 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts...
I0420 07:29:04.010731 134410976347904 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts.
I0420 07:29:04.010919 134540780734272 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5.
W0420 07:29:04.011049 134540780734272 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.046252 seconds. If this number is high, consider checkpointing less frequently.
I0420 07:29:04.012447 134540780734272 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 9
I0420 07:29:04.014513 134540780734272 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/9.
I0420 07:29:04.314061 134540780734272 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0420 07:29:04.314151 134540780734272 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
I0420 07:29:04.351079 134540780734272 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.037844s
I0420 07:29:04.351207 134540780734272 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 37.349 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.041301727294921875 s) (per-host)
I0420 07:29:04.351253 134540780734272 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.041356s (batch_requests_ready=0.001748s, total_serialization_initiated=0.039550s, others=0.000058s)
I0420 07:29:04.351362 134540780734272 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.045623s (all_items=0.000010s, per_item={'items': '0.00001025'}, temp_paths=0.045613)
I0420 07:29:04.352004 134540780734272 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/9.
I0420 07:29:04.352353 134410976347904 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 07:49:04.352300
I0420 07:29:04.354259 134540780734272 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0420 07:29:04.354564 134405719906048 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0420 07:29:04.354730 134540780734272 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776670115.964773, 'wait_for_prev_duration_secs': 28.046252489089966, 'time_between_consecutive_saves_sec': 23.242911338806152, 'checkpointer_blocking_start_time': 1776670144.0124853, 'checkpointer_blocking_duration_secs': 0.340015172958374, 'get_old_steps_start_time': 1776670144.3525176, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776670115.9629326, 'checkpoint_manager_blocking_duration_secs': 28.391764879226685}
I0420 07:29:04.354847 134540780734272 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0420 07:29:04.354890 134540780734272 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0420 07:29:09.499418 134388826236672 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260420_060028/linen_xpk_main_20260420_060028_16_zarr3_false/checkpoints/9/items/array_metadatas/process_4
I0420 07:29:46.309034 134410976347904 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.610 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.99910259246826 s) (per-host)
I0420 07:29:46.309135 134410976347904 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.956664s.
I0420 07:29:54.234055 134410976347904 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 49.881569s.
I0420 07:29:54.234379 134405719906048 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0420 07:29:54.234506 134405719906048 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0420 07:29:54.234556 134405719906048 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0420 07:29:54.237572 134405719906048 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0420 07:29:54.237753 134540780734272 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0420 07:29:54.237904 134540780734272 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0420 07:29:54.238873 134540780734272 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 92.056, Tokens/s/device: 13875.621, total_weights: 65536, loss: 7.181, lm_loss: 7.181, perplexity: 1314.495
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Mon Apr 20 07:30:06 UTC 2026
EXIT_CODE=0
NNX  ·  574ad3fb9  ·  main_20260420_060028  ·  full log
XPK Start: Mon Apr 20 08:12:04 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-20 08:12:28.688246: 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)
I0420 08:12:29.223512 133873139013440 max_utils.py:273] Attempting to initialize the jax distributed system...
I0420 08:12:38.265778 133873139013440 distributed.py:149] Starting JAX distributed service on [::]:8482
I0420 08:12:38.268115 133873139013440 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-sv-96as5-slice-job-0-0.mt-16-zarr3-false-sv-96as5:8482
I0420 08:12:39.584028 133873139013440 max_utils.py:284] Jax distributed system initialized!
I0420 08:12:45.842317 133873139013440 max_utils.py:800] System Information: Jax Version: 0.9.2
I0420 08:12:45.842434 133873139013440 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0420 08:12:45.842473 133873139013440 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
I0420 08:12:45.842507 133873139013440 train_utils.py:348] 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 217, 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: Mon Apr 20 08:12:55 UTC 2026
EXIT_CODE=1