MaxView

‹ 06_grad_accumCase: 07_eval08_checkpoint_async_true_resume ›

Metrics: Linen vs NNX  ·  main

MetricLinen  7f7822822NNX  7f7822822Diff (NNX − Linen)
Parameters1.104 billion
Final loss8.1790
TFLOP/s92.138
Tok/s13888.0
Avg s/step2.637
Memory %1.38
JAX0.9.20.9.2

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

Linen  ·  7f7822822  ·  main_20260417_125726  ·  full log
XPK Start: Fri Apr 17 13:31:07 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 13:31:32.079266: 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)
I0417 13:31:32.255441 139231935276864 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 13:31:41.294741 139231935276864 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 13:31:41.297156 139231935276864 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-n3c99-slice-job-0-0.mt-07-eval-n3c99:8482
I0417 13:31:42.357735 139231935276864 max_utils.py:284] Jax distributed system initialized!
I0417 13:31:48.750384 139231935276864 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 13:31:48.750484 139231935276864 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 13:31:48.750523 139231935276864 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
I0417 13:31:48.750556 139231935276864 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 13:31:49.427139 139231935276864 maxtext_utils.py:1548] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 13:31:49.427407 139231935276864 checkpointing.py:677] Setting up checkpoint logger...
I0417 13:31:49.427460 139231935276864 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 13:31:49.427503 139231935276864 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 13:31:49.427852 139231935276864 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 0x7ea0be44c2f0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 13:31:53.044189 139231935276864 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0417 13:31:53.044414 139231935276864 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e8d143f80b0>}, handler_registry=None
I0417 13:31:53.044646 139231935276864 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e8d143f80b0>` 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`.
I0417 13:31:53.044694 139231935276864 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e8d143f9700>` 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`.
I0417 13:31:53.044729 139231935276864 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 0x7e8d143f80b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e8d143f80b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e8d143f9700>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e8d143f9700>}).
I0417 13:31:53.045069 139231935276864 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 13:31:53.045141 139231935276864 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e8cf46b89a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 13:31:55.172669 139231935276864 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints
I0417 13:31:55.607766 139231935276864 checkpoint_manager.py:929] [process=0][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_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e8d145f2ed0>
I0417 13:31:55.607981 139231935276864 checkpointing.py:301] Checkpoint manager created!
I0417 13:31:56.515180 139231935276864 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 13:31:56.515293 139231935276864 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:31:56.889580 139231935276864 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:31:56.889666 139231935276864 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:31:56.905605 139231935276864 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 13:31:56.905663 139231935276864 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 13:31:56.929146 139231935276864 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:31:56.929213 139231935276864 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:31:56.945364 139231935276864 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:31:56.945430 139231935276864 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:31:56.962052 139231935276864 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 13:31:56.962113 139231935276864 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:31:56.986905 139231935276864 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 13:31:56.986978 139231935276864 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 13:31:57.013972 139231935276864 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 13:31:57.014072 139231935276864 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 13:31:57.234101 139231935276864 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 13:31:57.234209 139231935276864 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0417 13:31:58.652577 139231935276864 maxtext_utils.py:1651]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 13:31:58.652699 139231935276864 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 13:31:58.652751 139231935276864 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 13:31:58.652807 139231935276864 maxtext_utils.py:1651]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   P('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 13:31:58.652874 139231935276864 maxtext_utils.py:1651]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 13:31:58.652914 139231935276864 maxtext_utils.py:1651]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 13:31:58.652966 139231935276864 maxtext_utils.py:1651]  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)
I0417 13:31:58.653017 139231935276864 maxtext_utils.py:1651]  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')
I0417 13:31:58.653055 139231935276864 maxtext_utils.py:1651]  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)
I0417 13:31:58.653091 139231935276864 maxtext_utils.py:1651]  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)
I0417 13:31:58.653137 139231935276864 maxtext_utils.py:1651]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0417 13:31:58.653189 139231935276864 maxtext_utils.py:1651]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0417 13:31:59.140950 139231935276864 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:31:59.141043 139231935276864 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:31:59.156210 139231935276864 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 13:31:59.156276 139231935276864 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 13:32:09.943321 139231935276864 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.
I0417 13:32:09.950944 139231935276864 metric_logger.py:301] number parameters: 1.104 billion
I0417 13:32:21.448204 139231935276864 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0417 13:32:21.601511 139231935276864 checkpointing.py:776] Waited 0.15328717231750488 seconds for step 0 to finish before starting checkpointing.
I0417 13:32:21.603987 139231935276864 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:32:21.605564 139231935276864 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0
I0417 13:32:21.607245 139231935276864 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0.
I0417 13:32:21.952784 139231935276864 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 13:32:21.966920 139231935276864 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 13:32:21.967033 139231935276864 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
I0417 13:32:22.209544 139105123952384 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0
I0417 13:32:22.235349 139231935276864 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.269332s
I0417 13:32:22.236296 139231935276864 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.592 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27583980560302734 s) (per-host)
I0417 13:32:22.236369 139231935276864 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.275932s (batch_requests_ready=0.002201s, total_serialization_initiated=0.272866s, others=0.000866s)
I0417 13:32:22.236452 139231935276864 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.279855s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.279837)
I0417 13:32:22.237251 139231935276864 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.63 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0.
I0417 13:32:22.237580 139105132345088 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 13:52:22.237544
I0417 13:32:22.257484 139231935276864 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:32:22.257796 139103474558720 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:32:22.257984 139231935276864 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432741.603969, 'wait_for_prev_duration_secs': 6.151199340820312e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432741.6056015, 'checkpointer_blocking_duration_secs': 0.6321213245391846, 'get_old_steps_start_time': 1776432742.2377477, 'get_old_steps_duration_secs': 2.765655517578125e-05, 'checkpoint_manager_blocking_start_time': 1776432741.601983, 'checkpoint_manager_blocking_duration_secs': 0.6559615135192871}
I0417 13:32:22.258096 139231935276864 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0417 13:32:22.258146 139231935276864 max_utils.py:750] 
Memstats: After params initialized:
I0417 13:32:22.258197 139231935276864 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0))
I0417 13:32:22.258230 139231935276864 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0))
I0417 13:32:22.258255 139231935276864 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0))
I0417 13:32:22.258279 139231935276864 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0))
I0417 13:32:22.599376 139231935276864 metric_logger.py:196] completed step: 0, seconds: 11.192, TFLOP/s/device: 1.214, Tokens/s/device: 182.980, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0417 13:32:22.602737 139231935276864 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/tensorboard/'
I0417 13:32:23.027496 139231935276864 metric_logger.py:196] completed step: 1, seconds: 1.149, TFLOP/s/device: 11.822, Tokens/s/device: 1781.953, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875
I0417 13:32:23.173631 139231935276864 metric_logger.py:196] completed step: 2, seconds: 0.429, TFLOP/s/device: 31.683, Tokens/s/device: 4775.551, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129
I0417 13:32:23.320914 139231935276864 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 1542.064, Tokens/s/device: 232436.727, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203
I0417 13:32:24.414539 139105107166976 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/items
I0417 13:32:24.559982 139102441879296 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432744145219751, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776432744145219751, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:32:26.009031    2844 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 13:32:28.230636 139104583464704 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/items/array_metadatas/process_0
I0417 13:32:44.874682 139231935276864 metric_logger.py:196] completed step: 4, seconds: 0.147, TFLOP/s/device: 92.243, Tokens/s/device: 13903.881, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614
I0417 13:32:44.883591 139231935276864 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.315, Tokens/s/device: 13914.650, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717
I0417 13:32:45.029657 139231935276864 metric_logger.py:196] completed step: 6, seconds: 21.553, TFLOP/s/device: 0.630, Tokens/s/device: 95.022, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759
I0417 13:32:45.177253 139231935276864 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2084.555, Tokens/s/device: 314206.812, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905
I0417 13:32:45.324585 139231935276864 metric_logger.py:196] completed step: 8, seconds: 0.147, TFLOP/s/device: 92.304, Tokens/s/device: 13913.043, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406
I0417 13:32:45.471226 139231935276864 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0417 13:32:45.471953 139231935276864 checkpointing.py:776] Waited 0.0007476806640625 seconds for step 9 to finish before starting checkpointing.
I0417 13:32:45.474092 139231935276864 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:33:03.076543 139104029239040 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.066923s (commit=34.969410s, array_metadata_write=2.097513s)
I0417 13:33:03.077650 139105132345088 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 38.417 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.1171658039093 s) (per-host)
I0417 13:33:03.077775 139105132345088 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.840085s.
I0417 13:33:04.286143 139105132345088 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432744145219751, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:33:05.508322 139105132345088 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000247s.
I0417 13:33:05.863696 139105132345088 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 13:33:05.878465 139102441879296 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776432744145219751, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/_CHECKPOINT_METADATA
I0417 13:33:06.797525 139105132345088 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.960238s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/items
I0417 13:33:06.799240 139105132345088 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0/items
I0417 13:33:07.339595 139105132345088 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0
I0417 13:33:09.612400 139105132345088 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0`.
I0417 13:33:09.613302 139105132345088 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 48.01 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/0
I0417 13:33:09.614717 139105132345088 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 47.377026s.
I0417 13:33:09.614929 139103474558720 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:33:09.615052 139103474558720 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:33:09.615117 139103474558720 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:33:09.618046 139103474558720 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0417 13:33:09.618188 139231935276864 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0417 13:33:09.618270 139231935276864 checkpoint_manager.py:1452] Waiting for previous save to complete took 24.144183 seconds. If this number is high, consider checkpointing less frequently.
I0417 13:33:09.620139 139231935276864 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9
I0417 13:33:09.622395 139231935276864 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9.
I0417 13:33:10.350063 139231935276864 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 13:33:10.350153 139231935276864 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
I0417 13:33:10.376042 139231935276864 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026683s
I0417 13:33:10.376853 139231935276864 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 49.561 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03112506866455078 s) (per-host)
I0417 13:33:10.376927 139231935276864 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031226s (batch_requests_ready=0.001832s, total_serialization_initiated=0.028663s, others=0.000731s)
I0417 13:33:10.377008 139231935276864 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035303s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.035287)
I0417 13:33:10.377727 139231935276864 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9.
I0417 13:33:10.378013 139103474558720 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 13:53:10.377989
I0417 13:33:10.392674 139231935276864 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 13:33:10.392949 139104033425152 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 13:33:10.393085 139231935276864 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776432765.474061, 'wait_for_prev_duration_secs': 24.144182920455933, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776432789.6201787, 'checkpointer_blocking_duration_secs': 0.7579343318939209, 'get_old_steps_start_time': 1776432790.3781376, 'get_old_steps_duration_secs': 3.337860107421875e-05, 'checkpoint_manager_blocking_start_time': 1776432765.4721844, 'checkpoint_manager_blocking_duration_secs': 24.92086672782898}
I0417 13:33:10.393194 139231935276864 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0417 13:33:10.393241 139231935276864 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 13:33:10.678748 139105132345088 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9
I0417 13:33:14.125968 139105107166976 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9/items
I0417 13:33:17.853318 139104050210560 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9/items/array_metadatas/process_0
I0417 13:33:57.191191 139104041817856 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.949851s (commit=40.712444s, array_metadata_write=1.237407s)
I0417 13:33:57.192300 139103474558720 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 33.719 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.84655404090881 s) (per-host)
I0417 13:33:57.192423 139103474558720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 46.814335s.
I0417 13:33:58.955793 139103474558720 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 13:33:59.080117 139103474558720 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000271s.
I0417 13:33:59.748442 139103474558720 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.687148s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9/items
I0417 13:33:59.750164 139103474558720 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9/items
I0417 13:34:00.272833 139103474558720 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9
I0417 13:34:01.691984 139103474558720 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9`.
I0417 13:34:01.692812 139103474558720 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.07 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_07_eval/checkpoints/9
I0417 13:34:01.694411 139103474558720 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.316321s.
I0417 13:34:01.694619 139104033425152 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 13:34:01.694740 139104033425152 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 13:34:01.694800 139104033425152 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0417 13:34:01.696344 139104033425152 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0417 13:34:01.696488 139231935276864 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0417 13:34:01.696578 139231935276864 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 13:34:01.697533 139231935276864 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.138, Tokens/s/device: 13888.041, 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 17 13:34:13 UTC 2026
EXIT_CODE=0
NNX  ·  7f7822822  ·  main_20260417_125726  ·  full log
XPK Start: Fri Apr 17 15:20:24 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 15:20:47.880081: 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)
I0417 15:20:48.059388 134203344451392 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 15:20:57.102164 134203344451392 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 15:20:57.104443 134203344451392 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-y7jvc-slice-job-0-0.mt-07-eval-y7jvc:8482
I0417 15:20:58.572693 134203344451392 max_utils.py:284] Jax distributed system initialized!
I0417 15:21:03.866632 134203344451392 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 15:21:03.866736 134203344451392 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 15:21:03.866778 134203344451392 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
I0417 15:21:03.866810 134203344451392 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 712, 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 708, in main
    train_func()
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 698, in train_func
    run(config, recorder, diagnostic_config)
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 677, in run
    train_loop(config, recorder)
  File "/deps/src/maxtext/trainers/pre_train/train.py", line 504, 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: Fri Apr 17 15:21:14 UTC 2026
EXIT_CODE=1