MaxView

XPK Start: Thu Apr 23 15:47:58 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 15:48:23.116605: 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 15:48:23.327202 138295921416000 max_utils.py:273] Attempting to initialize the jax distributed system...
I0423 15:48:32.368308 138295921416000 distributed.py:149] Starting JAX distributed service on [::]:8482
I0423 15:48:32.370753 138295921416000 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-ofr4a-slice-job-0-0.mt-16-zarr3-false-rs-ofr4a:8482
I0423 15:48:33.662167 138295921416000 max_utils.py:284] Jax distributed system initialized!
I0423 15:48:39.881335 138295921416000 max_utils.py:800] System Information: Jax Version: 0.9.2
I0423 15:48:39.881439 138295921416000 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0423 15:48:39.881484 138295921416000 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 15:48:39.881520 138295921416000 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0423 15:48:40.582420 138295921416000 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0423 15:48:40.694904 138295921416000 checkpointing.py:688] Setting up checkpoint logger...
I0423 15:48:40.695019 138295921416000 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False
I0423 15:48:40.695064 138295921416000 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0423 15:48:40.695281 138295921416000 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 0x7dc6f571fbf0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0423 15:48:42.371492 138295921416000 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0423 15:48:42.371745 138295921416000 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 0x7db28c469c10>}, handler_registry=None
I0423 15:48:42.372014 138295921416000 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db28c469c10>` 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 15:48:42.372087 138295921416000 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db2bc30f440>` 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 15:48:42.372152 138295921416000 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 0x7db28c469c10>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7db28c469c10>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db2bc30f440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7db2bc30f440>}).
I0423 15:48:42.372534 138295921416000 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0423 15:48:42.372623 138295921416000 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7db28c4d7560> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0423 15:48:43.827242 138174500157184 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776959060421050359, 'commit_timestamp_nsecs': 1776959107587462092, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA
I0423 15:48:44.064119 138174500157184 checkpoint_manager.py:1871] Missing metrics for step 0
I0423 15:48:44.110109 138295921416000 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints
I0423 15:48:44.635815 138295921416000 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=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/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7db2bc305610>
I0423 15:48:44.635995 138295921416000 checkpointing.py:302] Checkpoint manager created!
I0423 15:48:45.615891 138295921416000 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0423 15:48:45.616013 138295921416000 checkpointing.py:582] restoring from this run's directory step 9
I0423 15:48:45.867762 138295921416000 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9.
I0423 15:48:46.114367 138295921416000 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9.
I0423 15:48:46.114810 138295921416000 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9.
I0423 15:48:47.531884    1459 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0423 15:48:54.730512 138295921416000 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 219.328 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.205072402954102 s) (per-host)
I0423 15:48:54.739785 138295921416000 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 207.744 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.618062734603882 s) (per-host)
I0423 15:48:55.275653 138295921416000 event_tracking.py:138] [process=6] [sync] Finished load in 9.41 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9
I0423 15:48:55.275810 138295921416000 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776959325.8676634, 'checkpointer_duration_secs': 9.408114194869995, 'checkpoint_manager_start_time': 1776959325.6166246, 'checkpoint_manager_duration_secs': 9.65915560722351}
fsdp: 32
I0423 15:48:55.351297 138295921416000 maxtext_utils.py:1889]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0423 15:48:55.351423 138295921416000 maxtext_utils.py:1889]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0423 15:48:55.351508 138295921416000 maxtext_utils.py:1889]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0423 15:48:55.351598 138295921416000 maxtext_utils.py:1889]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0423 15:48:55.351644 138295921416000 maxtext_utils.py:1889]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0423 15:48:55.351680 138295921416000 maxtext_utils.py:1889]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0423 15:48:55.351716 138295921416000 maxtext_utils.py:1889]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0423 15:48:55.351750 138295921416000 maxtext_utils.py:1889]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0423 15:48:55.351784 138295921416000 maxtext_utils.py:1889]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0423 15:48:55.351816 138295921416000 maxtext_utils.py:1889]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0423 15:48:55.351849 138295921416000 maxtext_utils.py:1889]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)

I0423 15:48:55.351880 138295921416000 maxtext_utils.py:1889]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')
I0423 15:48:55.754171 138295921416000 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 15:48:55.754265 138295921416000 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 15:48:55.759665 138295921416000 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0423 15:48:55.759722 138295921416000 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 15:48:55.776198 138295921416000 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 15:48:55.776260 138295921416000 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 15:48:55.792110 138295921416000 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0423 15:48:55.792168 138295921416000 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0423 15:48:55.816153 138295921416000 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 15:48:55.816215 138295921416000 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 15:48:55.832190 138295921416000 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 15:48:55.832247 138295921416000 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 15:48:55.848063 138295921416000 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 15:48:55.848134 138295921416000 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 15:48:55.878684 138295921416000 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0423 15:48:55.878756 138295921416000 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0423 15:48:55.898758 138295921416000 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0423 15:48:55.898819 138295921416000 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0423 15:49:07.449030 138295921416000 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 15:49:07.451336 138295921416000 metric_logger.py:301] number parameters: 1.104 billion
I0423 15:49:18.768929 138295921416000 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0423 15:49:18.924144 138295921416000 checkpointing.py:798] Waited 0.15518784523010254 seconds for step 10 to finish before starting checkpointing.
I0423 15:49:18.926700 138295921416000 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 15:49:18.928693 138295921416000 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0423 15:49:18.930197 138295921416000 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/10.
I0423 15:49:19.271683 138295921416000 signaling_client.py:364] Using JaxDistributedSignalingClient
I0423 15:49:19.272684 138295921416000 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0423 15:49:19.272740 138295921416000 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 15:49:19.551990 138295921416000 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.281084s
I0423 15:49:19.552173 138295921416000 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.353 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2881476879119873 s) (per-host)
I0423 15:49:19.552223 138295921416000 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288207s (batch_requests_ready=0.002804s, total_serialization_initiated=0.285336s, others=0.000067s)
I0423 15:49:19.552317 138295921416000 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292187s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.292169)
I0423 15:49:19.552986 138295921416000 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/10.
I0423 15:49:19.553338 138172805650176 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:09:19.553300
I0423 15:49:19.558727 138295921416000 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 15:49:19.559017 138170197530368 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 15:49:19.559158 138295921416000 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959358.9266806, 'wait_for_prev_duration_secs': 6.437301635742188e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776959358.9287324, 'checkpointer_blocking_duration_secs': 0.6247615814208984, 'get_old_steps_start_time': 1776959359.5535197, 'get_old_steps_duration_secs': 3.337860107421875e-05, 'checkpoint_manager_blocking_start_time': 1776959358.9245002, 'checkpoint_manager_blocking_duration_secs': 0.6346228122711182}
I0423 15:49:19.559324 138295921416000 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0423 15:49:19.559402 138295921416000 max_utils.py:750] 
Memstats: After params initialized:
I0423 15:49:19.559451 138295921416000 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0423 15:49:19.559486 138295921416000 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0423 15:49:19.559509 138295921416000 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0423 15:49:19.559530 138295921416000 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0423 15:49:19.873920 138295921416000 metric_logger.py:196] completed step: 10, seconds: 11.316, TFLOP/s/device: 1.201, Tokens/s/device: 180.975, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537
I0423 15:49:20.059897 138295921416000 metric_logger.py:196] completed step: 11, seconds: 1.103, TFLOP/s/device: 12.316, Tokens/s/device: 1856.371, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821
I0423 15:49:20.207703 138295921416000 metric_logger.py:196] completed step: 12, seconds: 0.029, TFLOP/s/device: 465.648, Tokens/s/device: 70187.464, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013
I0423 15:49:20.355307 138295921416000 metric_logger.py:196] completed step: 13, seconds: 0.165, TFLOP/s/device: 82.460, Tokens/s/device: 12429.221, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520
I0423 15:49:20.491400 138295921416000 checkpointing.py:794] Waiting for step 15 to finish before checkpoint...
I0423 15:49:20.784154 138295921416000 checkpointing.py:798] Waited 0.2927076816558838 seconds for step 15 to finish before starting checkpointing.
I0423 15:49:20.786933 138295921416000 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 15:49:24.852296 138170205923072 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6
I0423 15:50:01.656687 138172805650176 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.261 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.39262890815735 s) (per-host)
I0423 15:50:01.656796 138172805650176 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.103343s.
I0423 15:50:08.699682 138172805650176 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.146212s.
I0423 15:50:08.699973 138170197530368 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 15:50:08.700090 138170197530368 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 15:50:08.700377 138170197530368 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0423 15:50:08.702616 138170197530368 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0423 15:50:08.702792 138295921416000 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0423 15:50:08.702924 138295921416000 checkpoint_manager.py:1452] Waiting for previous save to complete took 47.915990 seconds. If this number is high, consider checkpointing less frequently.
I0423 15:50:08.704817 138295921416000 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15
I0423 15:50:08.707364 138295921416000 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/15.
I0423 15:50:09.007421 138295921416000 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0423 15:50:09.007516 138295921416000 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 15:50:09.047693 138295921416000 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041831s
I0423 15:50:09.047860 138295921416000 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.714 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04864025115966797 s) (per-host)
I0423 15:50:09.047908 138295921416000 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048696s (batch_requests_ready=0.002565s, total_serialization_initiated=0.046067s, others=0.000063s)
I0423 15:50:09.047997 138295921416000 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053716s (all_items=0.000013s, per_item={'items': '0.00001287'}, temp_paths=0.053703)
I0423 15:50:09.048672 138295921416000 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/15.
I0423 15:50:09.049016 138170197530368 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:10:09.048979
I0423 15:50:09.058458 138295921416000 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 15:50:09.058763 138167998990080 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 15:50:09.058922 138295921416000 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959360.7869036, 'wait_for_prev_duration_secs': 47.915990114212036, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776959408.7048552, 'checkpointer_blocking_duration_secs': 0.3442859649658203, 'get_old_steps_start_time': 1776959409.049167, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1776959360.7845218, 'checkpoint_manager_blocking_duration_secs': 48.27436375617981}
I0423 15:50:09.059037 138295921416000 checkpointing.py:409] Started an asynchronous checkpoint save for step 15
I0423 15:50:09.060070 138295921416000 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.214, Tokens/s/device: 13899.541, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067
I0423 15:50:09.067987 138295921416000 metric_logger.py:196] completed step: 15, seconds: 0.278, TFLOP/s/device: 48.956, Tokens/s/device: 7379.196, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780
I0423 15:50:14.267786 138170205923072 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6
I0423 15:50:30.965343 138295921416000 metric_logger.py:196] completed step: 16, seconds: 48.575, TFLOP/s/device: 0.280, Tokens/s/device: 42.162, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201
I0423 15:50:31.112979 138295921416000 metric_logger.py:196] completed step: 17, seconds: 21.752, TFLOP/s/device: 0.625, Tokens/s/device: 94.151, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443
I0423 15:50:31.260446 138295921416000 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.979, Tokens/s/device: 13411.831, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909
I0423 15:50:31.408773 138295921416000 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0423 15:50:31.409447 138295921416000 checkpointing.py:798] Waited 0.0006864070892333984 seconds for step 19 to finish before starting checkpointing.
I0423 15:50:31.411700 138295921416000 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 15:50:53.206653 138170197530368 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 35.732 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 44.20739197731018 s) (per-host)
I0423 15:50:53.206781 138170197530368 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.157108s.
I0423 15:50:58.660382 138170197530368 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.610693s.
I0423 15:50:58.660694 138167998990080 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 15:50:58.660808 138167998990080 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 15:50:58.660857 138167998990080 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0423 15:50:58.663071 138167998990080 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0423 15:50:58.663333 138295921416000 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15.
W0423 15:50:58.663465 138295921416000 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.251768 seconds. If this number is high, consider checkpointing less frequently.
I0423 15:50:58.664952 138295921416000 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0423 15:50:58.666963 138295921416000 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/19.
I0423 15:50:58.960561 138295921416000 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0423 15:50:58.960652 138295921416000 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 15:50:59.002564 138295921416000 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043569s
I0423 15:50:59.002718 138295921416000 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 30.610 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.050394296646118164 s) (per-host)
I0423 15:50:59.002768 138295921416000 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.050454s (batch_requests_ready=0.002645s, total_serialization_initiated=0.047745s, others=0.000064s)
I0423 15:50:59.002861 138295921416000 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054632s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.054620)
I0423 15:50:59.003576 138295921416000 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/19.
I0423 15:50:59.003838 138167998990080 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:10:59.003814
I0423 15:50:59.005728 138295921416000 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0423 15:50:59.005942 138169047553792 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0423 15:50:59.006057 138295921416000 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959431.4116707, 'wait_for_prev_duration_secs': 27.251768350601196, 'time_between_consecutive_saves_sec': 22.709012746810913, 'checkpointer_blocking_start_time': 1776959458.6649935, 'checkpointer_blocking_duration_secs': 0.33893847465515137, 'get_old_steps_start_time': 1776959459.0039513, 'get_old_steps_duration_secs': 2.6702880859375e-05, 'checkpoint_manager_blocking_start_time': 1776959431.409684, 'checkpoint_manager_blocking_duration_secs': 27.596341609954834}
I0423 15:50:59.006186 138295921416000 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0423 15:50:59.006230 138295921416000 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0423 15:51:04.110528 138170205923072 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6
I0423 15:51:41.230146 138167998990080 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.362 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.27776861190796 s) (per-host)
I0423 15:51:41.230255 138167998990080 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.226346s.
I0423 15:51:48.156584 138167998990080 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.152658s.
I0423 15:51:48.156805 138169047553792 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0423 15:51:48.156872 138169047553792 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0423 15:51:48.156922 138169047553792 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0423 15:51:48.159281 138169047553792 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0423 15:51:48.159451 138295921416000 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0423 15:51:48.159626 138295921416000 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0423 15:51:48.160592 138295921416000 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 91.865, Tokens/s/device: 13846.820, total_weights: 65536, loss: 5.385, lm_loss: 5.385, perplexity: 218.045
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Thu Apr 23 15:51:59 UTC 2026
EXIT_CODE=0