MaxView

XPK Start: Sat Apr 25 14:59:33 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. 
2026-04-25 14:59:58.685068: 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)
I0425 14:59:58.897606 135722684299072 max_utils.py:273] Attempting to initialize the jax distributed system...
I0425 15:00:07.937961 135722684299072 distributed.py:149] Starting JAX distributed service on [::]:8482
I0425 15:00:07.940295 135722684299072 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-7pio8-slice-job-0-0.mt-16-zarr3-false-rs-7pio8:8482
I0425 15:00:08.990527 135722684299072 max_utils.py:284] Jax distributed system initialized!
I0425 15:00:15.224218 135722684299072 max_utils.py:800] System Information: Jax Version: 0.9.2
I0425 15:00:15.224323 135722684299072 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0425 15:00:15.224366 135722684299072 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
I0425 15:00:15.224401 135722684299072 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0425 15:00:15.923363 135722684299072 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0425 15:00:16.035680 135722684299072 checkpointing.py:688] Setting up checkpoint logger...
I0425 15:00:16.035800 135722684299072 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False
I0425 15:00:16.035845 135722684299072 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0425 15:00:16.036049 135722684299072 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 0x7b6fdf18eba0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0425 15:00:17.320953 135722684299072 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0425 15:00:17.321214 135722684299072 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 0x7b5b3c63deb0>}, handler_registry=None
I0425 15:00:17.321465 135722684299072 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b5b3c63deb0>` 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`.
I0425 15:00:17.321515 135722684299072 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b5b3c63df40>` 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`.
I0425 15:00:17.321550 135722684299072 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 0x7b5b3c63deb0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b5b3c63deb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b5b3c63df40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b5b3c63df40>}).
I0425 15:00:17.321869 135722684299072 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0425 15:00:17.321938 135722684299072 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b5b3c6336a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0425 15:00:18.834638 135600237704960 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777128955382748983, 'commit_timestamp_nsecs': 1777129000616370604, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA
I0425 15:00:19.088229 135600237704960 checkpoint_manager.py:1871] Missing metrics for step 0
I0425 15:00:19.115852 135722684299072 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints
I0425 15:00:19.528707 135722684299072 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b6fc817a990>
I0425 15:00:19.528874 135722684299072 checkpointing.py:302] Checkpoint manager created!
I0425 15:00:20.498453 135722684299072 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0425 15:00:20.498577 135722684299072 checkpointing.py:582] restoring from this run's directory step 9
I0425 15:00:20.762249 135722684299072 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9.
I0425 15:00:21.008374 135722684299072 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9.
I0425 15:00:21.008806 135722684299072 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9.
I0425 15:00:22.428434    1469 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0425 15:00:30.091745 135722684299072 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 206.037 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.669865608215332 s) (per-host)
I0425 15:00:30.100377 135722684299072 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 195.903 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 8.078530550003052 s) (per-host)
I0425 15:00:30.833455 135722684299072 event_tracking.py:138] [process=6] [sync] Finished load in 10.07 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9
I0425 15:00:30.833617 135722684299072 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1777129220.7621489, 'checkpointer_duration_secs': 10.071436882019043, 'checkpoint_manager_start_time': 1777129220.4992378, 'checkpoint_manager_duration_secs': 10.334349870681763}
fsdp: 32
I0425 15:00:30.909052 135722684299072 maxtext_utils.py:1889]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0425 15:00:30.909190 135722684299072 maxtext_utils.py:1889]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0425 15:00:30.909248 135722684299072 maxtext_utils.py:1889]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0425 15:00:30.909288 135722684299072 maxtext_utils.py:1889]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0425 15:00:30.909326 135722684299072 maxtext_utils.py:1889]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0425 15:00:30.909358 135722684299072 maxtext_utils.py:1889]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0425 15:00:30.909394 135722684299072 maxtext_utils.py:1889]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0425 15:00:30.909428 135722684299072 maxtext_utils.py:1889]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0425 15:00:30.909461 135722684299072 maxtext_utils.py:1889]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0425 15:00:30.909493 135722684299072 maxtext_utils.py:1889]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0425 15:00:30.909524 135722684299072 maxtext_utils.py:1889]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)
I0425 15:00:30.909562 135722684299072 maxtext_utils.py:1889]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')

I0425 15:00:31.312123 135722684299072 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0425 15:00:31.312216 135722684299072 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0425 15:00:31.317608 135722684299072 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0425 15:00:31.317665 135722684299072 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0425 15:00:31.334429 135722684299072 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn').
I0425 15:00:31.334488 135722684299072 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0425 15:00:31.350831 135722684299072 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn').
I0425 15:00:31.350894 135722684299072 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0425 15:00:31.377085 135722684299072 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0425 15:00:31.377175 135722684299072 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0425 15:00:31.393341 135722684299072 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0425 15:00:31.393402 135722684299072 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0425 15:00:31.409399 135722684299072 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim').
I0425 15:00:31.409459 135722684299072 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0425 15:00:31.438884 135722684299072 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv').
I0425 15:00:31.438956 135722684299072 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0425 15:00:31.459035 135722684299072 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0425 15:00:31.459119 135722684299072 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0425 15:00:42.966611 135722684299072 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.
I0425 15:00:42.969003 135722684299072 metric_logger.py:301] number parameters: 1.104 billion
I0425 15:00:54.234057 135722684299072 checkpointing.py:794] Waiting for step 10 to finish before checkpoint...
I0425 15:00:54.386053 135722684299072 checkpointing.py:798] Waited 0.15198349952697754 seconds for step 10 to finish before starting checkpointing.
I0425 15:00:54.388751 135722684299072 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0425 15:00:54.390789 135722684299072 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10
I0425 15:00:54.392215 135722684299072 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10.
I0425 15:00:54.747653 135722684299072 signaling_client.py:364] Using JaxDistributedSignalingClient
I0425 15:00:54.748668 135722684299072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0425 15:00:54.748737 135722684299072 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
I0425 15:00:55.027663 135722684299072 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280800s
I0425 15:00:55.027866 135722684299072 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.362 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2877047061920166 s) (per-host)
I0425 15:00:55.027929 135722684299072 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287784s (batch_requests_ready=0.002539s, total_serialization_initiated=0.285154s, others=0.000091s)
I0425 15:00:55.028047 135722684299072 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292659s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.292642)
I0425 15:00:55.028834 135722684299072 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.64 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10.
I0425 15:00:55.029157 135599633725184 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 15:20:55.029125
I0425 15:00:55.035964 135722684299072 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0425 15:00:55.036280 135596395714304 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0425 15:00:55.036441 135722684299072 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777129254.388733, 'wait_for_prev_duration_secs': 6.079673767089844e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777129254.3908274, 'checkpointer_blocking_duration_secs': 0.6384859085083008, 'get_old_steps_start_time': 1777129255.0293393, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1777129254.3864493, 'checkpoint_manager_blocking_duration_secs': 0.6499567031860352}
I0425 15:00:55.036554 135722684299072 checkpointing.py:409] Started an asynchronous checkpoint save for step 10
I0425 15:00:55.036626 135722684299072 max_utils.py:750] 
Memstats: After params initialized:
I0425 15:00:55.036673 135722684299072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0))
I0425 15:00:55.036705 135722684299072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0))
I0425 15:00:55.036730 135722684299072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0))
I0425 15:00:55.036754 135722684299072 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0))
I0425 15:00:55.353399 135722684299072 metric_logger.py:196] completed step: 10, seconds: 11.264, TFLOP/s/device: 1.206, Tokens/s/device: 181.819, total_weights: 65536, loss: 6.827, lm_loss: 6.827, perplexity: 922.537
I0425 15:00:55.533646 135722684299072 metric_logger.py:196] completed step: 11, seconds: 1.118, TFLOP/s/device: 12.157, Tokens/s/device: 1832.503, total_weights: 65536, loss: 6.522, lm_loss: 6.522, perplexity: 679.821
I0425 15:00:55.681402 135722684299072 metric_logger.py:196] completed step: 12, seconds: 0.023, TFLOP/s/device: 590.770, Tokens/s/device: 89047.350, total_weights: 65536, loss: 6.260, lm_loss: 6.260, perplexity: 523.013
I0425 15:00:55.829483 135722684299072 metric_logger.py:196] completed step: 13, seconds: 0.165, TFLOP/s/device: 82.181, Tokens/s/device: 12387.272, total_weights: 65536, loss: 6.039, lm_loss: 6.039, perplexity: 419.520
I0425 15:00:55.835984 135722684299072 checkpointing.py:794] Waiting for step 15 to finish before checkpoint...
I0425 15:00:56.124441 135722684299072 checkpointing.py:798] Waited 0.28842759132385254 seconds for step 15 to finish before starting checkpointing.
I0425 15:00:56.127240 135722684299072 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0425 15:00:59.953258 135598023112448 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6
I0425 15:01:36.347072 135599633725184 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.965 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.60689067840576 s) (per-host)
I0425 15:01:36.347195 135599633725184 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.317922s.
I0425 15:01:44.241899 135599633725184 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.212611s.
I0425 15:01:44.242217 135596395714304 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0425 15:01:44.242341 135596395714304 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0425 15:01:44.242390 135596395714304 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0425 15:01:44.244002 135596395714304 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0425 15:01:44.244205 135722684299072 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.
W0425 15:01:44.244346 135722684299072 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.117106 seconds. If this number is high, consider checkpointing less frequently.
I0425 15:01:44.246297 135722684299072 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15
I0425 15:01:44.248780 135722684299072 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15.
I0425 15:01:44.559806 135722684299072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0425 15:01:44.559898 135722684299072 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
I0425 15:01:44.585050 135722684299072 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026881s
I0425 15:01:44.585189 135722684299072 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 45.765 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0337064266204834 s) (per-host)
I0425 15:01:44.585238 135722684299072 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.033764s (batch_requests_ready=0.002572s, total_serialization_initiated=0.031129s, others=0.000063s)
I0425 15:01:44.585314 135722684299072 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.038633s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.038621)
I0425 15:01:44.586041 135722684299072 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15.
I0425 15:01:44.586371 135596395714304 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 15:21:44.586339
I0425 15:01:44.610496 135722684299072 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0425 15:01:44.610764 135599465936640 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0425 15:01:44.610898 135722684299072 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777129256.1272101, 'wait_for_prev_duration_secs': 48.117106437683105, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777129304.246337, 'checkpointer_blocking_duration_secs': 0.3401782512664795, 'get_old_steps_start_time': 1777129304.5865421, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1777129256.1248107, 'checkpoint_manager_blocking_duration_secs': 48.48605394363403}
I0425 15:01:44.611011 135722684299072 checkpointing.py:409] Started an asynchronous checkpoint save for step 15
I0425 15:01:44.612218 135722684299072 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.201, Tokens/s/device: 13897.465, total_weights: 65536, loss: 5.858, lm_loss: 5.858, perplexity: 350.067
I0425 15:01:44.620388 135722684299072 metric_logger.py:196] completed step: 15, seconds: 0.148, TFLOP/s/device: 91.577, Tokens/s/device: 13803.423, total_weights: 65536, loss: 5.713, lm_loss: 5.713, perplexity: 302.780
I0425 15:01:49.756937 135593694574336 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6
I0425 15:02:06.857923 135722684299072 metric_logger.py:196] completed step: 16, seconds: 48.783, TFLOP/s/device: 0.279, Tokens/s/device: 41.982, total_weights: 65536, loss: 5.599, lm_loss: 5.599, perplexity: 270.201
I0425 15:02:07.005158 135722684299072 metric_logger.py:196] completed step: 17, seconds: 22.092, TFLOP/s/device: 0.615, Tokens/s/device: 92.701, total_weights: 65536, loss: 5.511, lm_loss: 5.511, perplexity: 247.443
I0425 15:02:07.152540 135722684299072 metric_logger.py:196] completed step: 18, seconds: 0.153, TFLOP/s/device: 88.648, Tokens/s/device: 13362.041, total_weights: 65536, loss: 5.442, lm_loss: 5.442, perplexity: 230.909
I0425 15:02:07.300627 135722684299072 checkpointing.py:794] Waiting for step 19 to finish before checkpoint...
I0425 15:02:07.301328 135722684299072 checkpointing.py:798] Waited 0.0007119178771972656 seconds for step 19 to finish before starting checkpointing.
I0425 15:02:07.303461 135722684299072 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0425 15:02:26.986866 135596395714304 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.224 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.43534708023071 s) (per-host)
I0425 15:02:26.986978 135596395714304 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.400497s.
I0425 15:02:33.260872 135596395714304 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.674376s.
I0425 15:02:33.261168 135599465936640 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0425 15:02:33.261308 135599465936640 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0425 15:02:33.261358 135599465936640 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0425 15:02:33.262946 135599465936640 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0425 15:02:33.263133 135722684299072 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.
W0425 15:02:33.263259 135722684299072 checkpoint_manager.py:1452] Waiting for previous save to complete took 25.959805 seconds. If this number is high, consider checkpointing less frequently.
I0425 15:02:33.264756 135722684299072 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19
I0425 15:02:33.266854 135722684299072 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19.
I0425 15:02:33.582303 135722684299072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array.
I0425 15:02:33.582398 135722684299072 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
I0425 15:02:33.608938 135722684299072 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.028328s
I0425 15:02:33.609132 135722684299072 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 43.750 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03525900840759277 s) (per-host)
I0425 15:02:33.609187 135722684299072 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035344s (batch_requests_ready=0.002675s, total_serialization_initiated=0.032579s, others=0.000090s)
I0425 15:02:33.609295 135722684299072 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.039517s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.039507)
I0425 15:02:33.609929 135722684299072 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19.
I0425 15:02:33.610218 135593694574336 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 15:22:33.610183
I0425 15:02:33.614772 135722684299072 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0425 15:02:33.615041 135599465936640 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0425 15:02:33.615215 135722684299072 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777129327.3034306, 'wait_for_prev_duration_secs': 25.95980453491211, 'time_between_consecutive_saves_sec': 23.059385061264038, 'checkpointer_blocking_start_time': 1777129353.264797, 'checkpointer_blocking_duration_secs': 0.34555745124816895, 'get_old_steps_start_time': 1777129353.6103706, 'get_old_steps_duration_secs': 2.2172927856445312e-05, 'checkpoint_manager_blocking_start_time': 1777129327.301565, 'checkpoint_manager_blocking_duration_secs': 26.313619136810303}
I0425 15:02:33.615315 135722684299072 checkpointing.py:409] Started an asynchronous checkpoint save for step 19
I0425 15:02:33.615360 135722684299072 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0425 15:02:38.941524 135593728145152 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6
I0425 15:03:16.625033 135593694574336 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.691 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.05114269256592 s) (per-host)
I0425 15:03:16.625177 135593694574336 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.014852s.
I0425 15:03:21.670047 135593694574336 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.059705s.
I0425 15:03:21.670346 135599465936640 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0425 15:03:21.670477 135599465936640 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0425 15:03:21.670520 135599465936640 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0425 15:03:21.672172 135599465936640 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0425 15:03:21.672367 135722684299072 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.
I0425 15:03:21.672541 135722684299072 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0425 15:03:21.673485 135722684299072 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.117, Tokens/s/device: 13884.840, 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: Sat Apr 25 15:03:33 UTC 2026
EXIT_CODE=0