XPK Start: Sat Apr 25 08:02:10 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 08:02:35.082081: 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 08:02:35.289055 136968299673408 max_utils.py:273] Attempting to initialize the jax distributed system... I0425 08:02:44.331565 136968299673408 distributed.py:149] Starting JAX distributed service on [::]:8482 I0425 08:02:44.333987 136968299673408 distributed.py:172] Connecting to JAX distributed service on mt-14-async-ckpt-false--ho85w-slice-job-0-0.mt-14-async-ckpt-false--ho85w:8482 I0425 08:02:45.217814 136968299673408 max_utils.py:284] Jax distributed system initialized! I0425 08:02:50.203523 136968299673408 max_utils.py:800] System Information: Jax Version: 0.9.2 I0425 08:02:50.203627 136968299673408 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0425 08:02:50.203669 136968299673408 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 08:02:50.203703 136968299673408 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 08:02:50.897987 136968299673408 maxtext_utils.py:1604] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1) I0425 08:02:50.898296 136968299673408 checkpointing.py:677] Setting up checkpoint logger... I0425 08:02:50.898353 136968299673408 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0425 08:02:50.898398 136968299673408 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0425 08:02:50.898751 136968299673408 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 0x7c91c9d1c080>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 08:02:52.223928 136968299673408 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0425 08:02:52.224186 136968299673408 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 0x7c7d6064c560>}, handler_registry=None I0425 08:02:52.224435 136968299673408 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c7d6064c560>` 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 08:02:52.224484 136968299673408 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7d6064fb00>` 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 08:02:52.224526 136968299673408 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 0x7c7d6064c560>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c7d6064c560>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7d6064fb00>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c7d6064fb00>}). I0425 08:02:52.224857 136968299673408 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0425 08:02:53.675961 136850232551168 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777103881767797599, 'commit_timestamp_nsecs': 1777103924944501731, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/0/_CHECKPOINT_METADATA I0425 08:02:53.930001 136850232551168 checkpoint_manager.py:1871] Missing metrics for step 0 I0425 08:02:53.953696 136968299673408 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints I0425 08:02:54.433919 136968299673408 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=False, 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_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c7d607aa6c0> I0425 08:02:54.434137 136968299673408 checkpointing.py:301] Checkpoint manager created! I0425 08:02:55.373572 136968299673408 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 08:02:55.373687 136968299673408 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 08:02:55.755105 136968299673408 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 08:02:55.755198 136968299673408 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 08:02:55.771872 136968299673408 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 08:02:55.771933 136968299673408 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 08:02:55.795998 136968299673408 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 08:02:55.796068 136968299673408 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 08:02:55.813006 136968299673408 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 08:02:55.813069 136968299673408 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 08:02:55.829751 136968299673408 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 08:02:55.829813 136968299673408 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 08:02:55.854799 136968299673408 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0425 08:02:55.854871 136968299673408 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 08:02:55.876310 136968299673408 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 08:02:55.876378 136968299673408 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 08:02:56.086198 136968299673408 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0425 08:02:56.086309 136968299673408 checkpointing.py:581] restoring from this run's directory step 9 I0425 08:02:56.325705 136968299673408 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/9. I0425 08:02:56.881784 136968299673408 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/9. I0425 08:02:56.882243 136968299673408 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.56 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/9. I0425 08:02:58.276228 1407 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 08:03:05.861376 136968299673408 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 208.174 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.591135740280151 s) (per-host) I0425 08:03:05.867004 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 197.673 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 8.006137371063232 s) (per-host) I0425 08:03:06.630150 136968299673408 event_tracking.py:138] [process=6] [sync] Finished load in 10.30 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/9 I0425 08:03:06.630311 136968299673408 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints', 'checkpointer_start_time': 1777104176.325625, 'checkpointer_duration_secs': 10.304653882980347, 'checkpoint_manager_start_time': 1777104176.086612, 'checkpoint_manager_duration_secs': 10.543668985366821} fsdp: 32 I0425 08:03:07.048473 136968299673408 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0425 08:03:07.048604 136968299673408 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 08:03:07.048658 136968299673408 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 08:03:07.048716 136968299673408 maxtext_utils.py:1707] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0425 08:03:07.048770 136968299673408 maxtext_utils.py:1707] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 08:03:07.048809 136968299673408 maxtext_utils.py:1707] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 08:03:07.048860 136968299673408 maxtext_utils.py:1707] 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) I0425 08:03:07.048911 136968299673408 maxtext_utils.py:1707] 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') I0425 08:03:07.048950 136968299673408 maxtext_utils.py:1707] 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) I0425 08:03:07.048985 136968299673408 maxtext_utils.py:1707] 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) I0425 08:03:07.049031 136968299673408 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0425 08:03:07.049076 136968299673408 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0425 08:03:07.535433 136968299673408 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 08:03:07.535525 136968299673408 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0425 08:03:07.551298 136968299673408 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 08:03:07.551359 136968299673408 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0425 08:03:18.668524 136968299673408 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 08:03:18.669392 136968299673408 metric_logger.py:301] number parameters: 1.104 billion I0425 08:03:29.868676 136968299673408 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0425 08:03:30.178912 136968299673408 checkpointing.py:776] Waited 0.3102118968963623 seconds for step 10 to finish before starting checkpointing. I0425 08:03:30.181334 136968299673408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 08:03:30.183317 136968299673408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0425 08:03:30.184918 136968299673408 event_tracking.py:70] [process=6] [sync] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/10. I0425 08:03:32.674807 136968299673408 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 08:03:32.676158 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:03:34.102117 136968299673408 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 08:03:34.102218 136968299673408 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 08:03:34.384160 136968299673408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.283098s I0425 08:03:34.384351 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.365 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28752875328063965 s) (per-host) I0425 08:03:34.384407 136968299673408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.287596s (batch_requests_ready=0.002469s, total_serialization_initiated=0.285052s, others=0.000075s) I0425 08:03:34.391609 136968299673408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.716879s (all_items=0.000033s, per_item={'items': '0.00003338'}, temp_paths=1.716846) I0425 08:03:34.391684 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:03:34.391733 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=1] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:03:36.390564 136849628571392 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/10/items/array_metadatas/process_6 I0425 08:04:12.937456 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 40.669 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.84061050415039 s) (per-host) I0425 08:04:18.362364 136968299673408 event_tracking.py:125] [process=6] [sync] Finished blocking save in 48.18 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/10. I0425 08:04:22.550919 136968299673408 event_tracking.py:138] [process=6] [sync] Finished save in 52.37 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/10 I0425 08:04:22.553555 136968299673408 checkpoint_manager.py:2137] [process=6][thread=MainThread][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0425 08:04:22.554941 136968299673408 checkpoint_manager.py:2146] [process=6][thread=MainThread][step=10] CheckpointManager Save Finalize is done on all hosts. I0425 08:04:22.554994 136968299673408 checkpoint_manager.py:1581] [process=6][thread=MainThread][step=10] Finished synchronous save. I0425 08:04:22.555047 136968299673408 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1777104210.1813157, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777104210.183358, 'checkpointer_blocking_duration_secs': 52.36768817901611, 'get_old_steps_start_time': 1777104262.5510683, 'get_old_steps_duration_secs': 4.4345855712890625e-05, 'checkpoint_manager_blocking_start_time': 1777104210.1794486, 'checkpoint_manager_blocking_duration_secs': 52.375571966171265} I0425 08:04:22.555146 136968299673408 checkpointing.py:410] Saved a checkpoint at step 10. I0425 08:04:22.555191 136968299673408 max_utils.py:750] Memstats: After params initialized: I0425 08:04:22.555237 136968299673408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0425 08:04:22.555267 136968299673408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0425 08:04:22.555292 136968299673408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0425 08:04:22.555316 136968299673408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0425 08:04:22.887272 136968299673408 metric_logger.py:196] completed step: 10, seconds: 11.199, TFLOP/s/device: 1.213, Tokens/s/device: 182.871, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0425 08:04:23.032920 136968299673408 metric_logger.py:196] completed step: 11, seconds: 52.998, TFLOP/s/device: 0.256, Tokens/s/device: 38.643, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0425 08:04:23.180259 136968299673408 metric_logger.py:196] completed step: 12, seconds: 0.024, TFLOP/s/device: 567.715, Tokens/s/device: 85572.222, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0425 08:04:23.327775 136968299673408 metric_logger.py:196] completed step: 13, seconds: 0.146, TFLOP/s/device: 92.869, Tokens/s/device: 13998.155, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0425 08:04:23.332417 136968299673408 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0425 08:04:23.621930 136968299673408 checkpointing.py:776] Waited 0.2894735336303711 seconds for step 15 to finish before starting checkpointing. I0425 08:04:23.624830 136968299673408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 08:04:23.626415 136968299673408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0425 08:04:23.628246 136968299673408 event_tracking.py:70] [process=6] [sync] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/15. I0425 08:04:26.048064 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:04:27.431605 136968299673408 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 08:04:27.431704 136968299673408 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 08:04:27.478476 136968299673408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.047975s I0425 08:04:27.478652 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 29.675 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05198216438293457 s) (per-host) I0425 08:04:27.478705 136968299673408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.052045s (batch_requests_ready=0.001984s, total_serialization_initiated=0.049990s, others=0.000071s) I0425 08:04:27.478804 136968299673408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.431991s (all_items=0.000025s, per_item={'items': '0.00002503'}, temp_paths=1.431966) I0425 08:04:27.478846 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:04:27.478883 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=2] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:04:29.500150 136849628571392 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/15/items/array_metadatas/process_6 I0425 08:05:03.918315 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 43.287 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.49162554740906 s) (per-host) I0425 08:05:10.423870 136968299673408 event_tracking.py:125] [process=6] [sync] Finished blocking save in 46.80 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/15. I0425 08:05:14.681008 136968299673408 event_tracking.py:138] [process=6] [sync] Finished save in 51.05 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/15 I0425 08:05:14.682858 136968299673408 checkpoint_manager.py:2137] [process=6][thread=MainThread][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0425 08:05:14.684639 136968299673408 checkpoint_manager.py:2146] [process=6][thread=MainThread][step=15] CheckpointManager Save Finalize is done on all hosts. I0425 08:05:14.684691 136968299673408 checkpoint_manager.py:1581] [process=6][thread=MainThread][step=15] Finished synchronous save. I0425 08:05:14.684740 136968299673408 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1777104263.6248088, 'wait_for_prev_duration_secs': 6.29425048828125e-05, 'time_between_consecutive_saves_sec': 1.0698261260986328, 'checkpointer_blocking_start_time': 1777104263.626456, 'checkpointer_blocking_duration_secs': 51.05469870567322, 'get_old_steps_start_time': 1777104314.6811767, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1777104263.6225984, 'checkpoint_manager_blocking_duration_secs': 51.06211757659912} I0425 08:05:14.684826 136968299673408 checkpointing.py:410] Saved a checkpoint at step 15. I0425 08:05:14.685649 136968299673408 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.168, Tokens/s/device: 13892.563, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0425 08:05:14.709304 136968299673408 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.122, Tokens/s/device: 13885.593, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0425 08:05:35.520880 136968299673408 metric_logger.py:196] completed step: 16, seconds: 51.357, TFLOP/s/device: 0.265, Tokens/s/device: 39.877, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0425 08:05:35.668356 136968299673408 metric_logger.py:196] completed step: 17, seconds: 20.683, TFLOP/s/device: 0.657, Tokens/s/device: 99.016, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0425 08:05:35.815708 136968299673408 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.411, Tokens/s/device: 13477.053, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0425 08:05:35.962412 136968299673408 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0425 08:05:35.963086 136968299673408 checkpointing.py:776] Waited 0.0006959438323974609 seconds for step 19 to finish before starting checkpointing. I0425 08:05:35.965229 136968299673408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 08:05:35.967209 136968299673408 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0425 08:05:35.968644 136968299673408 event_tracking.py:70] [process=6] [sync] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/19. I0425 08:05:38.463332 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:05:39.897588 136968299673408 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 08:05:39.897686 136968299673408 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 08:05:39.941788 136968299673408 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045251s I0425 08:05:39.941957 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.424 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.049088478088378906 s) (per-host) I0425 08:05:39.942007 136968299673408 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049147s (batch_requests_ready=0.001962s, total_serialization_initiated=0.047121s, others=0.000064s) I0425 08:05:39.948728 136968299673408 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 1.486737s (all_items=0.000023s, per_item={'items': '0.00002337'}, temp_paths=1.486713) I0425 08:05:39.948796 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:05:39.948841 136968299673408 future.py:372] [process=6][thread=MainThread][operation_id=3] _SignalingThread.join() waiting for signals ([]) blocking the main thread will slow down blocking save times. This is likely due to main thread calling result() on a CommitFuture. I0425 08:05:42.006287 136849628571392 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/19/items/array_metadatas/process_6 I0425 08:06:17.118052 136968299673408 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 42.434 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.225143909454346 s) (per-host) I0425 08:06:22.743560 136968299673408 event_tracking.py:125] [process=6] [sync] Finished blocking save in 46.78 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/19. I0425 08:06:26.249941 136968299673408 event_tracking.py:138] [process=6] [sync] Finished save in 50.28 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints/19 I0425 08:06:26.252838 136968299673408 checkpoint_manager.py:2137] [process=6][thread=MainThread][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0425 08:06:26.254546 136968299673408 checkpoint_manager.py:2146] [process=6][thread=MainThread][step=19] CheckpointManager Save Finalize is done on all hosts. I0425 08:06:26.254595 136968299673408 checkpoint_manager.py:1581] [process=6][thread=MainThread][step=19] Finished synchronous save. I0425 08:06:26.254644 136968299673408 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260425_071456/linen_xpk_main_20260425_071456_14_async_ckpt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': True, 'wait_for_prev_start_time': 1777104335.9652138, 'wait_for_prev_duration_secs': 6.127357482910156e-05, 'time_between_consecutive_saves_sec': 21.28053331375122, 'checkpointer_blocking_start_time': 1777104335.9672482, 'checkpointer_blocking_duration_secs': 50.28281211853027, 'get_old_steps_start_time': 1777104386.2500753, 'get_old_steps_duration_secs': 5.626678466796875e-05, 'checkpoint_manager_blocking_start_time': 1777104335.9633408, 'checkpoint_manager_blocking_duration_secs': 50.29128074645996} I0425 08:06:26.254750 136968299673408 checkpointing.py:410] Saved a checkpoint at step 19. I0425 08:06:26.254789 136968299673408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 08:06:26.254826 136968299673408 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 08:06:26.255567 136968299673408 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.061, Tokens/s/device: 13876.373, 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: Sat Apr 25 08:06:35 UTC 2026 EXIT_CODE=0