XPK Start: Tue Apr 21 15:09:11 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.
2026-04-21 15:09:36.186744: 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)
I0421 15:09:36.399779 135376663914304 max_utils.py:273] Attempting to initialize the jax distributed system...
I0421 15:09:45.442287 135376663914304 distributed.py:149] Starting JAX distributed service on [::]:8482
I0421 15:09:45.444769 135376663914304 distributed.py:172] Connecting to JAX distributed service on mt-08-checkpoint-async--dlki2-slice-job-0-0.mt-08-checkpoint-async--dlki2:8482
I0421 15:09:46.501400 135376663914304 max_utils.py:284] Jax distributed system initialized!
I0421 15:09:52.732577 135376663914304 max_utils.py:800] System Information: Jax Version: 0.9.2
I0421 15:09:52.732702 135376663914304 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0421 15:09:52.732744 135376663914304 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
I0421 15:09:52.732782 135376663914304 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0421 15:09:53.441556 135376663914304 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0421 15:09:53.441850 135376663914304 checkpointing.py:677] Setting up checkpoint logger...
I0421 15:09:53.441905 135376663914304 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0421 15:09:53.441951 135376663914304 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0421 15:09:53.442282 135376663914304 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 0x7b1f1e447740>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0421 15:09:54.707555 135376663914304 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0421 15:09:54.707810 135376663914304 checkpoint_manager.py:708] [process=4][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b0aa87a7500>}, handler_registry=None
I0421 15:09:54.708042 135376663914304 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b0aa87a7500>` 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`.
I0421 15:09:54.708089 135376663914304 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b0adbfdb6e0>` 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`.
I0421 15:09:54.708125 135376663914304 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 0x7b0aa87a7500>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b0aa87a7500>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b0adbfdb6e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b0adbfdb6e0>}).
I0421 15:09:54.708455 135376663914304 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0421 15:09:54.708523 135376663914304 async_checkpointer.py:192] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b0ac81b1580> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0421 15:09:57.433131 135254006290176 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776783936302952480, 'commit_timestamp_nsecs': 1776783977776104029, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/0/_CHECKPOINT_METADATA
I0421 15:09:57.759977 135254006290176 checkpoint_manager.py:1871] Missing metrics for step 0
I0421 15:09:58.159189 135376663914304 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints
I0421 15:09:58.870543 135376663914304 checkpoint_manager.py:929] [process=4][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b0ae00861b0>
I0421 15:09:58.870728 135376663914304 checkpointing.py:301] Checkpoint manager created!
I0421 15:09:59.803724 135376663914304 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0421 15:09:59.803841 135376663914304 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 15:10:00.183543 135376663914304 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 15:10:00.183647 135376663914304 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 15:10:00.200170 135376663914304 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 15:10:00.200226 135376663914304 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 15:10:00.224237 135376663914304 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 15:10:00.224313 135376663914304 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 15:10:00.240891 135376663914304 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 15:10:00.240954 135376663914304 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 15:10:00.257415 135376663914304 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0421 15:10:00.257475 135376663914304 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 15:10:00.282323 135376663914304 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0421 15:10:00.282394 135376663914304 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 15:10:00.304460 135376663914304 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0421 15:10:00.304525 135376663914304 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0421 15:10:00.520184 135376663914304 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0421 15:10:00.520299 135376663914304 checkpointing.py:581] restoring from this run's directory step 9
I0421 15:10:00.771162 135376663914304 event_tracking.py:70] [process=4] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/9.
I0421 15:10:01.032383 135376663914304 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/9.
I0421 15:10:01.032840 135376663914304 event_tracking.py:125] [process=4] [sync] Finished blocking load in 0.26 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/9.
I0421 15:10:02.453006 1471 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0421 15:10:08.995388 135376663914304 jax_array_handlers.py:843] [process=4] /jax/orbax/read/worker/io/requested throughput: 241.323 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.548375606536865 s) (per-host)
I0421 15:10:08.999994 135376663914304 base_pytree_checkpoint_handler.py:130] [process=4] /jax/checkpoint/read/gbytes_per_sec: 227.366 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.960543155670166 s) (per-host)
I0421 15:10:09.933106 135376663914304 event_tracking.py:138] [process=4] [sync] Finished load in 9.16 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/9
I0421 15:10:09.933266 135376663914304 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints', 'checkpointer_start_time': 1776784200.771062, 'checkpointer_duration_secs': 9.162171840667725, 'checkpoint_manager_start_time': 1776784200.5205414, 'checkpoint_manager_duration_secs': 9.412694454193115}
fsdp: 32
I0421 15:10:10.355182 135376663914304 maxtext_utils.py:1668] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0421 15:10:10.355316 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0421 15:10:10.355370 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0421 15:10:10.355429 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0421 15:10:10.355482 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0421 15:10:10.355524 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0421 15:10:10.355575 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/key/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0421 15:10:10.355626 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/out/kernel
Shape: float32[16,16,128,2048]
Logical: P('heads', 'layers', 'kv', 'embed')
Physical: (None, None, None, 'fsdp')
I0421 15:10:10.355680 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/query/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'q_heads', 'kv')
Physical: ('fsdp', None, None, None)
I0421 15:10:10.355718 135376663914304 maxtext_utils.py:1668] params/params/decoder/layers/self_attention/value/kernel
Shape: float32[2048,16,16,128]
Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim')
Physical: ('fsdp', None, None, None)
I0421 15:10:10.355763 135376663914304 maxtext_utils.py:1668] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0421 15:10:10.355810 135376663914304 maxtext_utils.py:1668] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0421 15:10:10.859380 135376663914304 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0421 15:10:10.859477 135376663914304 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0421 15:10:10.875091 135376663914304 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0421 15:10:10.875148 135376663914304 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0421 15:10:22.085848 135376663914304 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.
I0421 15:10:22.086750 135376663914304 metric_logger.py:301] number parameters: 1.104 billion
I0421 15:10:33.768044 135376663914304 checkpointing.py:772] Waiting for step 10 to finish before checkpoint...
I0421 15:10:33.921584 135376663914304 checkpointing.py:776] Waited 0.1535205841064453 seconds for step 10 to finish before starting checkpointing.
I0421 15:10:33.924178 135376663914304 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 15:10:33.926143 135376663914304 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 10
I0421 15:10:33.927712 135376663914304 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/10.
I0421 15:10:34.275941 135376663914304 signaling_client.py:364] Using JaxDistributedSignalingClient
I0421 15:10:34.276899 135376663914304 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0421 15:10:34.276954 135376663914304 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
I0421 15:10:34.558177 135376663914304 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.282374s
I0421 15:10:34.558353 135376663914304 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 5.385 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2864830493927002 s) (per-host)
I0421 15:10:34.558408 135376663914304 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.286549s (batch_requests_ready=0.002287s, total_serialization_initiated=0.284187s, others=0.000075s)
I0421 15:10:34.558498 135376663914304 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.290579s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.290562)
I0421 15:10:34.559225 135376663914304 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.63 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/10.
I0421 15:10:34.559584 135253419095808 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 15:30:34.559544
I0421 15:10:34.567746 135376663914304 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 15:10:34.568039 135250189477632 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0421 15:10:34.568193 135376663914304 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784233.924159, 'wait_for_prev_duration_secs': 6.222724914550781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784233.9261818, 'checkpointer_blocking_duration_secs': 0.6335704326629639, 'get_old_steps_start_time': 1776784234.5597765, 'get_old_steps_duration_secs': 3.075599670410156e-05, 'checkpoint_manager_blocking_start_time': 1776784233.9221365, 'checkpoint_manager_blocking_duration_secs': 0.646021842956543}
I0421 15:10:34.568300 135376663914304 checkpointing.py:408] Started an asynchronous checkpoint save for step 10
I0421 15:10:34.568354 135376663914304 max_utils.py:750]
Memstats: After params initialized:
I0421 15:10:34.568403 135376663914304 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0))
I0421 15:10:34.568435 135376663914304 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0))
I0421 15:10:34.568463 135376663914304 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0))
I0421 15:10:34.568488 135376663914304 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0))
I0421 15:10:34.884131 135376663914304 metric_logger.py:196] completed step: 10, seconds: 11.681, TFLOP/s/device: 1.163, Tokens/s/device: 175.325, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869
I0421 15:10:35.048462 135376663914304 metric_logger.py:196] completed step: 11, seconds: 1.114, TFLOP/s/device: 12.192, Tokens/s/device: 1837.712, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659
I0421 15:10:35.196295 135376663914304 metric_logger.py:196] completed step: 12, seconds: 0.017, TFLOP/s/device: 790.087, Tokens/s/device: 119090.539, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068
I0421 15:10:35.343762 135376663914304 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.760, Tokens/s/device: 13378.888, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001
I0421 15:10:35.348250 135376663914304 checkpointing.py:772] Waiting for step 15 to finish before checkpoint...
I0421 15:10:35.638254 135376663914304 checkpointing.py:776] Waited 0.28996825218200684 seconds for step 15 to finish before starting checkpointing.
I0421 15:10:35.641302 135376663914304 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 15:10:40.150174 135250197870336 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/10/items/array_metadatas/process_4
I0421 15:11:16.434900 135253419095808 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.464 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.16299104690552 s) (per-host)
I0421 15:11:16.435024 135253419095808 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.875312s.
I0421 15:11:25.769113 135253419095808 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 51.209384s.
I0421 15:11:25.769394 135250189477632 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 15:11:25.769511 135250189477632 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0421 15:11:25.769563 135250189477632 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts...
I0421 15:11:25.771244 135250189477632 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts.
I0421 15:11:25.771425 135376663914304 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10.
W0421 15:11:25.771556 135376663914304 checkpoint_manager.py:1452] Waiting for previous save to complete took 50.130256 seconds. If this number is high, consider checkpointing less frequently.
I0421 15:11:25.773476 135376663914304 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 15
I0421 15:11:25.776043 135376663914304 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/15.
I0421 15:11:26.515079 135376663914304 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0421 15:11:26.515173 135376663914304 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
I0421 15:11:26.555097 135376663914304 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041767s
I0421 15:11:26.555262 135376663914304 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 33.942 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04544782638549805 s) (per-host)
I0421 15:11:26.555310 135376663914304 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045505s (batch_requests_ready=0.001927s, total_serialization_initiated=0.043513s, others=0.000064s)
I0421 15:11:26.555399 135376663914304 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.050465s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.050453)
I0421 15:11:26.557308 135376663914304 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/15.
I0421 15:11:26.557672 135250189477632 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 15:31:26.557619
I0421 15:11:26.559519 135376663914304 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 15:11:26.559797 135250197870336 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0421 15:11:26.559960 135376663914304 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784235.6412716, 'wait_for_prev_duration_secs': 50.13025641441345, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776784285.773514, 'checkpointer_blocking_duration_secs': 0.7843048572540283, 'get_old_steps_start_time': 1776784286.557843, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776784235.638827, 'checkpoint_manager_blocking_duration_secs': 50.9211003780365}
I0421 15:11:26.560065 135376663914304 checkpointing.py:408] Started an asynchronous checkpoint save for step 15
I0421 15:11:26.560915 135376663914304 metric_logger.py:196] completed step: 14, seconds: 0.148, TFLOP/s/device: 91.869, Tokens/s/device: 13847.475, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193
I0421 15:11:26.566407 135376663914304 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.117, Tokens/s/device: 13884.840, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661
I0421 15:11:31.659183 135247513515776 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/15/items/array_metadatas/process_4
I0421 15:11:48.956181 135376663914304 metric_logger.py:196] completed step: 16, seconds: 51.217, TFLOP/s/device: 0.265, Tokens/s/device: 39.987, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872
I0421 15:11:49.103596 135376663914304 metric_logger.py:196] completed step: 17, seconds: 22.244, TFLOP/s/device: 0.611, Tokens/s/device: 92.071, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975
I0421 15:11:49.251072 135376663914304 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.469, Tokens/s/device: 13485.750, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333
I0421 15:11:49.397754 135376663914304 checkpointing.py:772] Waiting for step 19 to finish before checkpoint...
I0421 15:11:49.398411 135376663914304 checkpointing.py:776] Waited 0.0006742477416992188 seconds for step 19 to finish before starting checkpointing.
I0421 15:11:49.400569 135376663914304 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 15:12:09.005738 135250189477632 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.171 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.495882511138916 s) (per-host)
I0421 15:12:09.005861 135250189477632 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.448074s.
I0421 15:12:19.111226 135250189477632 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 52.553423s.
I0421 15:12:19.111496 135250197870336 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 15:12:19.111613 135250197870336 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0421 15:12:19.111684 135250197870336 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts...
I0421 15:12:19.116519 135250197870336 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts.
I0421 15:12:19.116722 135376663914304 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15.
W0421 15:12:19.116852 135376663914304 checkpoint_manager.py:1452] Waiting for previous save to complete took 29.716291 seconds. If this number is high, consider checkpointing less frequently.
I0421 15:12:19.118300 135376663914304 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 19
I0421 15:12:19.120358 135376663914304 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/19.
I0421 15:12:19.863492 135376663914304 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0421 15:12:19.863600 135376663914304 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
I0421 15:12:19.905945 135376663914304 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043468s
I0421 15:12:19.906096 135376663914304 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 32.701 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.047171831130981445 s) (per-host)
I0421 15:12:19.906143 135376663914304 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.047230s (batch_requests_ready=0.001779s, total_serialization_initiated=0.045389s, others=0.000063s)
I0421 15:12:19.906241 135376663914304 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051343s (all_items=0.000012s, per_item={'items': '0.00001240'}, temp_paths=0.051331)
I0421 15:12:19.906972 135376663914304 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.79 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/19.
I0421 15:12:19.907267 135250197870336 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 15:32:19.907235
I0421 15:12:19.909228 135376663914304 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 15:12:19.909467 135248587257600 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save.
I0421 15:12:19.909594 135376663914304 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776784309.4005373, 'wait_for_prev_duration_secs': 29.716291189193726, 'time_between_consecutive_saves_sec': 23.629252433776855, 'checkpointer_blocking_start_time': 1776784339.1183417, 'checkpointer_blocking_duration_secs': 0.7890622615814209, 'get_old_steps_start_time': 1776784339.9074225, 'get_old_steps_duration_secs': 2.4557113647460938e-05, 'checkpoint_manager_blocking_start_time': 1776784309.398668, 'checkpoint_manager_blocking_duration_secs': 30.510892152786255}
I0421 15:12:19.909726 135376663914304 checkpointing.py:408] Started an asynchronous checkpoint save for step 19
I0421 15:12:19.909770 135376663914304 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 15:12:25.505675 135247513515776 array_metadata_store.py:203] [process=4][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260421_144122/linen_xpk_main_20260421_144122_08_checkpoint_async_true/checkpoints/19/items/array_metadatas/process_4
I0421 15:13:01.687263 135250197870336 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.764 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.828295946121216 s) (per-host)
I0421 15:13:01.687380 135250197870336 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.780007s.
I0421 15:13:10.807939 135250197870336 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 50.900550s.
I0421 15:13:10.808249 135248587257600 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 15:13:10.808384 135248587257600 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save.
I0421 15:13:10.808427 135248587257600 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts...
I0421 15:13:10.809909 135248587257600 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts.
I0421 15:13:10.810093 135376663914304 checkpoint_manager.py:2032] [process=4][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19.
I0421 15:13:10.810269 135376663914304 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 15:13:10.811259 135376663914304 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.054, Tokens/s/device: 13875.339, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Tue Apr 21 15:13:22 UTC 2026
EXIT_CODE=0