XPK Start: Tue Apr 21 16:37:29 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-21 16:37:54.181339: 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 16:37:54.362270 132752748250944 max_utils.py:273] Attempting to initialize the jax distributed system...
I0421 16:38:03.402913 132752748250944 distributed.py:149] Starting JAX distributed service on [::]:8482
I0421 16:38:03.405277 132752748250944 distributed.py:172] Connecting to JAX distributed service on mt-04-int8-qvgta-slice-job-0-0.mt-04-int8-qvgta:8482
I0421 16:38:04.837455 132752748250944 max_utils.py:284] Jax distributed system initialized!
I0421 16:38:10.939127 132752748250944 max_utils.py:800] System Information: Jax Version: 0.9.2
I0421 16:38:10.939228 132752748250944 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0421 16:38:10.939269 132752748250944 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 16:38:10.939302 132752748250944 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0421 16:38:11.641343 132752748250944 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0421 16:38:11.641960 132752748250944 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0421 16:38:11.642259 132752748250944 checkpointing.py:688] Setting up checkpoint logger...
I0421 16:38:11.642314 132752748250944 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0421 16:38:11.642357 132752748250944 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0421 16:38:11.642699 132752748250944 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 0x78bc5798bf80>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0421 16:38:14.919220 132752748250944 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0421 16:38:14.919456 132752748250944 checkpoint_manager.py:708] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78a7e82781a0>}, handler_registry=None
I0421 16:38:14.919691 132752748250944 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78a7e82781a0>` 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 16:38:14.919738 132752748250944 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a7e827be60>` 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 16:38:14.919774 132752748250944 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 0x78a7e82781a0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78a7e82781a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a7e827be60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78a7e827be60>}).
I0421 16:38:14.920105 132752748250944 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0421 16:38:14.920175 132752748250944 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78a70c7782c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0421 16:38:15.822154 132752748250944 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints
I0421 16:38:15.848120 132752748250944 checkpoint_manager.py:929] [process=3][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78a7e846fb90>
I0421 16:38:15.848252 132752748250944 checkpointing.py:302] Checkpoint manager created!
I0421 16:38:16.998682 132752748250944 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0421 16:38:16.998789 132752748250944 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 16:38:17.387588 132752748250944 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 16:38:17.387678 132752748250944 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 16:38:17.404073 132752748250944 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0421 16:38:17.404130 132752748250944 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0421 16:38:17.469870 132752748250944 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 16:38:17.469965 132752748250944 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 16:38:17.486472 132752748250944 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 16:38:17.486530 132752748250944 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 16:38:17.503000 132752748250944 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 16:38:17.503059 132752748250944 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 16:38:17.528265 132752748250944 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0421 16:38:17.528332 132752748250944 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0421 16:38:17.598612 132752748250944 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0421 16:38:17.598698 132752748250944 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0421 16:38:18.114700 132752748250944 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0421 16:38:18.114809 132752748250944 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0421 16:38:20.585891 132752748250944 maxtext_utils.py:1821] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0421 16:38:20.586033 132752748250944 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_0/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0421 16:38:20.586085 132752748250944 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_1/kernel
Shape: float32[2048,16,7168]
Logical: P('embed', 'layers', 'mlp')
Physical: ('fsdp', None, None)
I0421 16:38:20.586152 132752748250944 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wo/kernel
Shape: float32[7168,16,2048]
Logical: P('mlp', 'layers', 'embed')
Physical: (None, None, 'fsdp')
I0421 16:38:20.586205 132752748250944 maxtext_utils.py:1821] params/params/decoder/layers/post_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0421 16:38:20.586241 132752748250944 maxtext_utils.py:1821] params/params/decoder/layers/pre_self_attention_layer_norm/scale
Shape: float32[2048,16]
Logical: P('norm', 'layers')
Physical: (None, None)
I0421 16:38:20.586293 132752748250944 maxtext_utils.py:1821] 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 16:38:20.586350 132752748250944 maxtext_utils.py:1821] 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 16:38:20.586392 132752748250944 maxtext_utils.py:1821] 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 16:38:20.586428 132752748250944 maxtext_utils.py:1821] 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 16:38:20.586481 132752748250944 maxtext_utils.py:1821] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0421 16:38:20.586528 132752748250944 maxtext_utils.py:1821] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0421 16:38:21.975862 132752748250944 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0421 16:38:21.975970 132752748250944 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0421 16:38:21.991605 132752748250944 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0421 16:38:21.991668 132752748250944 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0421 16:38:35.400247 132752748250944 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0421 16:38:35.401094 132752748250944 metric_logger.py:301] number parameters: 1.104 billion
I0421 16:38:50.580433 132752748250944 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0421 16:38:50.821602 132752748250944 checkpointing.py:798] Waited 0.2411513328552246 seconds for step 0 to finish before starting checkpointing.
I0421 16:38:50.825991 132752748250944 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 16:38:50.827335 132752748250944 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0
I0421 16:38:50.828750 132752748250944 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/0.
I0421 16:38:51.167741 132752748250944 signaling_client.py:364] Using JaxDistributedSignalingClient
I0421 16:38:51.168735 132752748250944 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0421 16:38:51.168792 132752748250944 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 16:38:51.447043 132752748250944 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.279376s
I0421 16:38:51.447218 132752748250944 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.410 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28514838218688965 s) (per-host)
I0421 16:38:51.447276 132752748250944 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285211s (batch_requests_ready=0.002321s, total_serialization_initiated=0.282820s, others=0.000071s)
I0421 16:38:51.447370 132752748250944 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289455s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.289438)
I0421 16:38:51.448152 132752748250944 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/0.
I0421 16:38:51.448431 132621429393152 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 16:58:51.448392
I0421 16:38:51.464640 132752748250944 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 16:38:51.464965 132619865945856 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save.
I0421 16:38:51.465121 132752748250944 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776789530.8259706, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776789530.8273752, 'checkpointer_blocking_duration_secs': 0.621204137802124, 'get_old_steps_start_time': 1776789531.4486022, 'get_old_steps_duration_secs': 2.7418136596679688e-05, 'checkpoint_manager_blocking_start_time': 1776789530.8221364, 'checkpoint_manager_blocking_duration_secs': 0.6429486274719238}
I0421 16:38:51.465227 132752748250944 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0421 16:38:51.465276 132752748250944 max_utils.py:750]
Memstats: After params initialized:
I0421 16:38:51.465323 132752748250944 max_utils.py:756] Using (GB) 0.44 / 31.25 (1.408000%) on TPU_10(process=3,(2,2,0,0))
I0421 16:38:51.465354 132752748250944 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0))
I0421 16:38:51.465380 132752748250944 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0))
I0421 16:38:51.465405 132752748250944 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0))
I0421 16:38:51.785660 132752748250944 metric_logger.py:196] completed step: 0, seconds: 15.179, TFLOP/s/device: 0.895, Tokens/s/device: 134.921, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277
I0421 16:38:52.054713 132752748250944 metric_logger.py:196] completed step: 1, seconds: 1.204, TFLOP/s/device: 11.288, Tokens/s/device: 1701.467, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277
I0421 16:38:52.486095 132752748250944 metric_logger.py:196] completed step: 2, seconds: 0.040, TFLOP/s/device: 338.890, Tokens/s/device: 51081.236, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28634.820
I0421 16:38:52.715313 132752748250944 metric_logger.py:196] completed step: 3, seconds: 0.432, TFLOP/s/device: 31.477, Tokens/s/device: 4744.530, total_weights: 65536, loss: 9.731, lm_loss: 9.731, perplexity: 16827.199
I0421 16:38:53.174518 132752748250944 metric_logger.py:196] completed step: 4, seconds: 0.235, TFLOP/s/device: 57.869, Tokens/s/device: 8722.688, total_weights: 65536, loss: 9.272, lm_loss: 9.272, perplexity: 10638.531
I0421 16:38:53.180690 132752748250944 metric_logger.py:196] completed step: 5, seconds: 0.229, TFLOP/s/device: 59.278, Tokens/s/device: 8935.038, total_weights: 65536, loss: 8.887, lm_loss: 8.887, perplexity: 7233.941
I0421 16:38:54.425181 2933 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0421 16:38:56.451572 132620344571648 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/0/items/array_metadatas/process_3
I0421 16:39:22.363350 132752748250944 metric_logger.py:196] completed step: 6, seconds: 0.460, TFLOP/s/device: 29.548, Tokens/s/device: 4453.723, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5365.468
I0421 16:39:22.592810 132752748250944 metric_logger.py:196] completed step: 7, seconds: 28.954, TFLOP/s/device: 0.469, Tokens/s/device: 70.732, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4358.779
I0421 16:39:22.822141 132752748250944 metric_logger.py:196] completed step: 8, seconds: 0.234, TFLOP/s/device: 58.092, Tokens/s/device: 8756.328, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.453
I0421 16:39:23.050977 132752748250944 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0421 16:39:23.051678 132752748250944 checkpointing.py:798] Waited 0.0007221698760986328 seconds for step 9 to finish before starting checkpointing.
I0421 16:39:23.053771 132752748250944 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 16:39:24.574122 132621429393152 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 47.276 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 33.412012815475464 s) (per-host)
I0421 16:39:24.574250 132621429393152 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 33.125708s.
I0421 16:39:32.332645 132621429393152 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 40.884087s.
I0421 16:39:32.332957 132619865945856 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 16:39:32.333079 132619865945856 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save.
I0421 16:39:32.333129 132619865945856 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0421 16:39:32.334446 132619865945856 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0421 16:39:32.334613 132752748250944 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0421 16:39:32.334743 132752748250944 checkpoint_manager.py:1452] Waiting for previous save to complete took 9.280971 seconds. If this number is high, consider checkpointing less frequently.
I0421 16:39:32.336614 132752748250944 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9
I0421 16:39:32.338729 132752748250944 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/9.
I0421 16:39:32.649890 132752748250944 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0421 16:39:32.649998 132752748250944 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 16:39:32.684425 132752748250944 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035580s
I0421 16:39:32.684586 132752748250944 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 39.405 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.039147138595581055 s) (per-host)
I0421 16:39:32.684636 132752748250944 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039206s (batch_requests_ready=0.001918s, total_serialization_initiated=0.037220s, others=0.000068s)
I0421 16:39:32.684727 132752748250944 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043312s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.043295)
I0421 16:39:32.685387 132752748250944 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/9.
I0421 16:39:32.685749 132619865945856 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 16:59:32.685711
I0421 16:39:32.687741 132752748250944 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0421 16:39:32.688049 132596688209664 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save.
I0421 16:39:32.688212 132752748250944 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776789563.0537407, 'wait_for_prev_duration_secs': 9.280970811843872, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776789572.3366523, 'checkpointer_blocking_duration_secs': 0.34923648834228516, 'get_old_steps_start_time': 1776789572.685912, 'get_old_steps_duration_secs': 4.887580871582031e-05, 'checkpoint_manager_blocking_start_time': 1776789563.051904, 'checkpoint_manager_blocking_duration_secs': 9.63627314567566}
I0421 16:39:32.688321 132752748250944 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0421 16:39:32.688365 132752748250944 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0421 16:39:37.790204 132596872668928 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900/linen_xpk_feat_nnx_trainstate_and_training_loop_20260421_161900_04_int8/checkpoints/9/items/array_metadatas/process_3
I0421 16:40:13.769620 132619865945856 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 38.410 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.12413454055786 s) (per-host)
I0421 16:40:13.769749 132619865945856 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.083889s.
I0421 16:40:23.186144 132619865945856 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.500269s.
I0421 16:40:23.186432 132596688209664 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save.
I0421 16:40:23.186545 132596688209664 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save.
I0421 16:40:23.186591 132596688209664 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0421 16:40:23.188730 132596688209664 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0421 16:40:23.188968 132752748250944 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0421 16:40:23.189126 132752748250944 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0421 16:40:23.190264 132752748250944 metric_logger.py:196] completed step: 9, seconds: 0.229, TFLOP/s/device: 59.236, Tokens/s/device: 8928.727, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3552.333
Per train step:
Total TFLOPs: 13.59
split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Tue Apr 21 16:40:35 UTC 2026
EXIT_CODE=0