XPK Start: Tue Apr 21 07:47:39 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 07:48:04.247378: 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 07:48:04.457487 140079299143488 max_utils.py:273] Attempting to initialize the jax distributed system... I0421 07:48:13.499676 140079299143488 distributed.py:149] Starting JAX distributed service on [::]:8482 I0421 07:48:13.502024 140079299143488 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-co9jn-slice-job-0-0.mt-16-zarr3-false-rs-co9jn:8482 I0421 07:48:14.386149 140079299143488 max_utils.py:284] Jax distributed system initialized! I0421 07:48:19.600774 140079299143488 max_utils.py:800] System Information: Jax Version: 0.9.2 I0421 07:48:19.600877 140079299143488 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0421 07:48:19.600918 140079299143488 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 07:48:19.600952 140079299143488 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 07:48:20.283356 140079299143488 maxtext_utils.py:1565] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 07:48:20.283632 140079299143488 checkpointing.py:677] Setting up checkpoint logger... I0421 07:48:20.283686 140079299143488 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=False I0421 07:48:20.283745 140079299143488 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0421 07:48:20.284082 140079299143488 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7f662096c6b0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 07:48:21.532498 140079299143488 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0421 07:48:21.532747 140079299143488 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 0x7f52886bdcd0>}, handler_registry=None I0421 07:48:21.532985 140079299143488 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f52886bdcd0>` 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 07:48:21.533033 140079299143488 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f52886c00e0>` 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 07:48:21.533068 140079299143488 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 0x7f52886bdcd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f52886bdcd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f52886c00e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f52886c00e0>}). I0421 07:48:21.533403 140079299143488 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0421 07:48:21.533473 140079299143488 async_checkpointer.py:192] [process=4][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7f52647a5800> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0421 07:48:23.571337 139960527070976 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776757537391658701, 'commit_timestamp_nsecs': 1776757584654464790, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/9/_CHECKPOINT_METADATA I0421 07:48:23.819859 139960527070976 checkpoint_manager.py:1871] Missing metrics for step 9 I0421 07:48:24.380169 140079299143488 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints I0421 07:48:25.069825 140079299143488 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_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f52886be870> I0421 07:48:25.069996 140079299143488 checkpointing.py:301] Checkpoint manager created! I0421 07:48:26.015293 140079299143488 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 07:48:26.015410 140079299143488 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 07:48:26.398523 140079299143488 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 07:48:26.398617 140079299143488 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 07:48:26.414863 140079299143488 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 07:48:26.414920 140079299143488 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 07:48:26.438578 140079299143488 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 07:48:26.438645 140079299143488 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 07:48:26.455153 140079299143488 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 07:48:26.455217 140079299143488 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 07:48:26.471542 140079299143488 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 07:48:26.471606 140079299143488 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 07:48:26.496201 140079299143488 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 07:48:26.496274 140079299143488 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 07:48:26.518155 140079299143488 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 07:48:26.518226 140079299143488 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 07:48:26.725754 140079299143488 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0421 07:48:26.725860 140079299143488 checkpointing.py:581] restoring from this run's directory step 9 I0421 07:48:26.975386 140079299143488 event_tracking.py:70] [process=4] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/9. I0421 07:48:27.222492 140079299143488 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/9. I0421 07:48:27.222942 140079299143488 event_tracking.py:125] [process=4] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/9. I0421 07:48:28.619527 1411 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 07:48:35.487646 140079299143488 jax_array_handlers.py:843] [process=4] /jax/orbax/read/worker/io/requested throughput: 229.890 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.874027729034424 s) (per-host) I0421 07:48:35.493254 140079299143488 base_pytree_checkpoint_handler.py:130] [process=4] /jax/checkpoint/read/gbytes_per_sec: 217.570 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 7.2739598751068115 s) (per-host) I0421 07:48:37.304486 140079299143488 event_tracking.py:138] [process=4] [sync] Finished load in 10.33 seconds @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/9 I0421 07:48:37.304637 140079299143488 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1776757706.9752903, 'checkpointer_duration_secs': 10.329314947128296, 'checkpoint_manager_start_time': 1776757706.726105, 'checkpoint_manager_duration_secs': 10.578502416610718} fsdp: 32 I0421 07:48:37.723950 140079299143488 maxtext_utils.py:1668] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0421 07:48:37.724075 140079299143488 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 07:48:37.724127 140079299143488 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 07:48:37.724186 140079299143488 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 07:48:37.724237 140079299143488 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 07:48:37.724285 140079299143488 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 07:48:37.724338 140079299143488 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 07:48:37.724392 140079299143488 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 07:48:37.724433 140079299143488 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 07:48:37.724469 140079299143488 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 07:48:37.724517 140079299143488 maxtext_utils.py:1668] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 07:48:37.724564 140079299143488 maxtext_utils.py:1668] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 07:48:38.209380 140079299143488 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 07:48:38.209475 140079299143488 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0421 07:48:38.224799 140079299143488 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 07:48:38.224860 140079299143488 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0421 07:48:49.102553 140079299143488 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 07:48:49.103338 140079299143488 metric_logger.py:301] number parameters: 1.104 billion I0421 07:49:00.788023 140079299143488 checkpointing.py:772] Waiting for step 10 to finish before checkpoint... I0421 07:49:01.169857 140079299143488 checkpointing.py:776] Waited 0.38181376457214355 seconds for step 10 to finish before starting checkpointing. I0421 07:49:01.172443 140079299143488 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 07:49:01.174353 140079299143488 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 10 I0421 07:49:01.176081 140079299143488 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/10. I0421 07:49:01.940504 140079299143488 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 07:49:01.941458 140079299143488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 07:49:01.941514 140079299143488 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 07:49:02.226637 140079299143488 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.286210s I0421 07:49:02.226828 140079299143488 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 5.316 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29015302658081055 s) (per-host) I0421 07:49:02.226882 140079299143488 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290217s (batch_requests_ready=0.002190s, total_serialization_initiated=0.287954s, others=0.000073s) I0421 07:49:02.226973 140079299143488 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294225s (all_items=0.000017s, per_item={'items': '0.00001693'}, temp_paths=0.294208) I0421 07:49:02.227740 140079299143488 event_tracking.py:125] [process=4] [async] Finished blocking save in 1.05 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/10. I0421 07:49:02.228078 139960132810496 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 08:09:02.228040 I0421 07:49:02.230071 140079299143488 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0421 07:49:02.230328 139958463481600 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 07:49:02.230455 140079299143488 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776757741.172425, 'wait_for_prev_duration_secs': 6.031990051269531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776757741.1743913, 'checkpointer_blocking_duration_secs': 1.0538363456726074, 'get_old_steps_start_time': 1776757742.2282524, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776757741.1703854, 'checkpoint_manager_blocking_duration_secs': 1.0600309371948242} I0421 07:49:02.230609 140079299143488 checkpointing.py:408] Started an asynchronous checkpoint save for step 10 I0421 07:49:02.230660 140079299143488 max_utils.py:750] Memstats: After params initialized: I0421 07:49:02.230719 140079299143488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_16(process=4,(0,4,0,0)) I0421 07:49:02.230753 140079299143488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_17(process=4,(1,4,0,0)) I0421 07:49:02.230779 140079299143488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_20(process=4,(0,5,0,0)) I0421 07:49:02.230803 140079299143488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_21(process=4,(1,5,0,0)) I0421 07:49:02.545122 140079299143488 metric_logger.py:196] completed step: 10, seconds: 11.685, TFLOP/s/device: 1.163, Tokens/s/device: 175.274, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0421 07:49:02.726032 140079299143488 metric_logger.py:196] completed step: 11, seconds: 1.756, TFLOP/s/device: 7.739, Tokens/s/device: 1166.579, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0421 07:49:02.873331 140079299143488 metric_logger.py:196] completed step: 12, seconds: 0.034, TFLOP/s/device: 401.713, Tokens/s/device: 60550.513, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0421 07:49:03.021070 140079299143488 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.705, Tokens/s/device: 13370.590, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0421 07:49:03.025404 140079299143488 checkpointing.py:772] Waiting for step 15 to finish before checkpoint... I0421 07:49:03.315321 140079299143488 checkpointing.py:776] Waited 0.28988027572631836 seconds for step 15 to finish before starting checkpointing. I0421 07:49:03.318346 140079299143488 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 07:49:07.181482 139958983567104 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_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/10/items/array_metadatas/process_4 I0421 07:49:43.336062 139960132810496 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 38.155 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.39935111999512 s) (per-host) I0421 07:49:43.336189 139960132810496 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.108001s. I0421 07:49:54.030608 139960132810496 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 51.802397s. I0421 07:49:54.030915 139958463481600 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 07:49:54.031039 139958463481600 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 07:49:54.031087 139958463481600 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0421 07:49:54.032935 139958463481600 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0421 07:49:54.033112 140079299143488 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 07:49:54.033245 140079299143488 checkpoint_manager.py:1452] Waiting for previous save to complete took 50.714901 seconds. If this number is high, consider checkpointing less frequently. I0421 07:49:54.035089 140079299143488 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 15 I0421 07:49:54.037326 140079299143488 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/15. I0421 07:49:54.327061 140079299143488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 07:49:54.327153 140079299143488 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 07:49:54.370073 140079299143488 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044886s I0421 07:49:54.370226 140079299143488 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 31.931 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.048309326171875 s) (per-host) I0421 07:49:54.370275 140079299143488 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048368s (batch_requests_ready=0.001769s, total_serialization_initiated=0.046534s, others=0.000065s) I0421 07:49:54.370361 140079299143488 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053132s (all_items=0.000012s, per_item={'items': '0.00001192'}, temp_paths=0.053120) I0421 07:49:54.371021 140079299143488 event_tracking.py:125] [process=4] [async] Finished blocking save in 0.34 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/15. I0421 07:49:54.371371 139958983567104 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 08:09:54.371333 I0421 07:49:54.373347 140079299143488 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0421 07:49:54.373645 139958463481600 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 07:49:54.373815 140079299143488 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776757743.3183162, 'wait_for_prev_duration_secs': 50.714900732040405, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776757794.0351279, 'checkpointer_blocking_duration_secs': 0.33638954162597656, 'get_old_steps_start_time': 1776757794.3715405, 'get_old_steps_duration_secs': 2.9087066650390625e-05, 'checkpoint_manager_blocking_start_time': 1776757743.3159277, 'checkpoint_manager_blocking_duration_secs': 51.05785274505615} I0421 07:49:54.373931 140079299143488 checkpointing.py:408] Started an asynchronous checkpoint save for step 15 I0421 07:49:54.374776 140079299143488 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.183, Tokens/s/device: 13894.825, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0421 07:49:54.380778 140079299143488 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.224, Tokens/s/device: 13900.956, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0421 07:50:00.462322 139953631643392 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_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/15/items/array_metadatas/process_4 I0421 07:50:15.780392 140079299143488 metric_logger.py:196] completed step: 16, seconds: 51.354, TFLOP/s/device: 0.265, Tokens/s/device: 39.880, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0421 07:50:15.927950 140079299143488 metric_logger.py:196] completed step: 17, seconds: 21.254, TFLOP/s/device: 0.639, Tokens/s/device: 96.360, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0421 07:50:16.075249 140079299143488 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.373, Tokens/s/device: 13471.203, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0421 07:50:16.221888 140079299143488 checkpointing.py:772] Waiting for step 19 to finish before checkpoint... I0421 07:50:16.222721 140079299143488 checkpointing.py:776] Waited 0.0008251667022705078 seconds for step 19 to finish before starting checkpointing. I0421 07:50:16.224712 140079299143488 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 07:50:36.145293 139958983567104 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.768 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.823349475860596 s) (per-host) I0421 07:50:36.145399 139958983567104 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.773912s. I0421 07:50:45.121665 139958983567104 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 50.750163s. I0421 07:50:45.121967 139958463481600 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 07:50:45.122087 139958463481600 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 07:50:45.122134 139958463481600 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0421 07:50:45.123847 139958463481600 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0421 07:50:45.124029 140079299143488 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 07:50:45.124158 140079299143488 checkpoint_manager.py:1452] Waiting for previous save to complete took 28.899465 seconds. If this number is high, consider checkpointing less frequently. I0421 07:50:45.125657 140079299143488 checkpoint_manager.py:1512] [process=4] Saving checkpoint at step 19 I0421 07:50:45.127566 140079299143488 event_tracking.py:70] [process=4] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/19. I0421 07:50:46.276094 140079299143488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 07:50:46.276195 140079299143488 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 07:50:46.320423 140079299143488 base_pytree_checkpoint_handler.py:154] [process=4][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045230s I0421 07:50:46.320593 140079299143488 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/blocking_gbytes_per_sec: 31.677 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04869675636291504 s) (per-host) I0421 07:50:46.320644 140079299143488 base_pytree_checkpoint_handler.py:768] [process=4][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048757s (batch_requests_ready=0.001625s, total_serialization_initiated=0.047067s, others=0.000065s) I0421 07:50:46.320754 140079299143488 composite_checkpoint_handler.py:715] [process=4][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.052886s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.052875) I0421 07:50:46.321432 140079299143488 event_tracking.py:125] [process=4] [async] Finished blocking save in 1.20 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/19. I0421 07:50:46.321787 139953631643392 async_checkpointer.py:76] [process=4][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 08:10:46.321750 I0421 07:50:46.323717 140079299143488 checkpoint_manager.py:1560] [process=4][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0421 07:50:46.323958 139958463481600 async_checkpointer.py:280] [process=4][thread=save_finalize] Waiting for background save thread=async_save. I0421 07:50:46.324120 140079299143488 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260421_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776757816.2246695, 'wait_for_prev_duration_secs': 28.899465084075928, 'time_between_consecutive_saves_sec': 22.191694259643555, 'checkpointer_blocking_start_time': 1776757845.125712, 'checkpointer_blocking_duration_secs': 1.196178674697876, 'get_old_steps_start_time': 1776757846.3219078, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776757816.222954, 'checkpoint_manager_blocking_duration_secs': 30.101131200790405} I0421 07:50:46.324255 140079299143488 checkpointing.py:408] Started an asynchronous checkpoint save for step 19 I0421 07:50:46.324309 140079299143488 checkpoint_manager.py:2020] [process=4][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 07:50:51.362440 139953614857984 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_061359/linen_xpk_main_20260421_061359_16_zarr3_false/checkpoints/19/items/array_metadatas/process_4 I0421 07:51:28.824223 139953631643392 base_pytree_checkpoint_handler.py:130] [process=4] /jax/orbax/write/gbytes_per_sec: 37.121 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.55228900909424 s) (per-host) I0421 07:51:28.824345 139953631643392 async_checkpointer.py:90] [process=4][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.501945s. I0421 07:51:37.648000 139953631643392 async_checkpointer.py:160] [process=4][thread=async_save] Background save thread done. Time taken: 51.325583s. I0421 07:51:37.648291 139958463481600 async_checkpointer.py:288] [process=4][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 07:51:37.648405 139958463481600 async_checkpointer.py:298] [process=4][thread=save_finalize] No errors found in background save thread=async_save. I0421 07:51:37.648444 139958463481600 checkpoint_manager.py:2137] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0421 07:51:37.650120 139958463481600 checkpoint_manager.py:2146] [process=4][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0421 07:51:37.650300 140079299143488 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 07:51:37.650457 140079299143488 checkpoint_manager.py:2009] [process=4][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 07:51:37.651288 140079299143488 metric_logger.py:196] completed step: 19, seconds: 0.148, TFLOP/s/device: 92.097, Tokens/s/device: 13881.922, 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 07:51:51 UTC 2026 EXIT_CODE=0