XPK Start: Fri Apr 24 14:48:07 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. 2026-04-24 14:48:32.026575: 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) I0424 14:48:32.240776 138027964458816 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:48:41.280212 138027964458816 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:48:41.282601 138027964458816 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-sv-ud3le-slice-job-0-0.mt-16-zarr3-false-sv-ud3le:8482 I0424 14:48:42.362565 138027964458816 max_utils.py:284] Jax distributed system initialized! I0424 14:48:48.462511 138027964458816 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:48:48.462619 138027964458816 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:48:48.462661 138027964458816 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 I0424 14:48:48.462701 138027964458816 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:48:49.174781 138027964458816 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:48:49.287872 138027964458816 checkpointing.py:688] Setting up checkpoint logger... I0424 14:48:49.288018 138027964458816 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0424 14:48:49.288066 138027964458816 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:48:49.288273 138027964458816 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 0x7d88a19a66c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:48:52.204098 138027964458816 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:48:52.204336 138027964458816 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 0x7d74086ba030>}, handler_registry=None I0424 14:48:52.204578 138027964458816 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d74086ba030>` 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`. I0424 14:48:52.204628 138027964458816 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d7448130e60>` 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`. I0424 14:48:52.204664 138027964458816 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 0x7d74086ba030>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d74086ba030>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d7448130e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d7448130e60>}). I0424 14:48:52.205021 138027964458816 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:48:52.205093 138027964458816 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d74086736a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:48:52.963407 138027964458816 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints I0424 14:48:52.968257 138027964458816 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=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d743038b8c0> I0424 14:48:52.968374 138027964458816 checkpointing.py:302] Checkpoint manager created! I0424 14:48:53.936576 138027964458816 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:48:53.936701 138027964458816 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 14:48:55.331559 138027964458816 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:48:55.331681 138027964458816 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:48:55.331733 138027964458816 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:48:55.331782 138027964458816 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:48:55.331820 138027964458816 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:48:55.331855 138027964458816 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:48:55.331892 138027964458816 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:48:55.331937 138027964458816 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:48:55.331977 138027964458816 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:48:55.332013 138027964458816 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:48:55.332047 138027964458816 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:48:55.332080 138027964458816 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:48:55.740253 138027964458816 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:48:55.740350 138027964458816 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:48:55.745620 138027964458816 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:48:55.745678 138027964458816 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:48:55.762154 138027964458816 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:48:55.762218 138027964458816 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:48:55.778027 138027964458816 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:48:55.778086 138027964458816 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:48:55.803597 138027964458816 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:48:55.803669 138027964458816 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:48:55.819629 138027964458816 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:48:55.819692 138027964458816 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:48:55.835551 138027964458816 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:48:55.835614 138027964458816 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:48:55.865189 138027964458816 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:48:55.865262 138027964458816 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:48:55.885181 138027964458816 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:48:55.885244 138027964458816 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:49:06.989043 138027964458816 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. I0424 14:49:06.991372 138027964458816 metric_logger.py:301] number parameters: 1.104 billion I0424 14:49:18.161613 138027964458816 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 14:49:18.333606 138027964458816 checkpointing.py:798] Waited 0.17197775840759277 seconds for step 0 to finish before starting checkpointing. I0424 14:49:18.336353 138027964458816 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:49:18.338085 138027964458816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0424 14:49:18.339689 138027964458816 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/0. I0424 14:49:18.635987 138027964458816 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:49:18.636966 138027964458816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:49:18.637025 138027964458816 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 I0424 14:49:18.914030 138027964458816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278806s I0424 14:49:18.914203 138027964458816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.357 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2879798412322998 s) (per-host) I0424 14:49:18.914255 138027964458816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.288043s (batch_requests_ready=0.003148s, total_serialization_initiated=0.284824s, others=0.000071s) I0424 14:49:18.914352 138027964458816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.292572s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.292554) I0424 14:49:18.915216 138027964458816 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.58 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/0. I0424 14:49:18.915529 137898380752640 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:09:18.915492 I0424 14:49:18.921631 138027964458816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:49:18.921855 137896732382976 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:49:18.921986 138027964458816 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042158.3363342, 'wait_for_prev_duration_secs': 6.198883056640625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777042158.338123, 'checkpointer_blocking_duration_secs': 0.5775516033172607, 'get_old_steps_start_time': 1777042158.9157019, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1777042158.3340838, 'checkpoint_manager_blocking_duration_secs': 0.587867021560669} I0424 14:49:18.922097 138027964458816 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 14:49:18.922166 138027964458816 max_utils.py:750] Memstats: After params initialized: I0424 14:49:18.922213 138027964458816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0424 14:49:18.922246 138027964458816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0424 14:49:18.922272 138027964458816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0424 14:49:18.922296 138027964458816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0424 14:49:19.235260 138027964458816 metric_logger.py:196] completed step: 0, seconds: 11.169, TFLOP/s/device: 1.216, Tokens/s/device: 183.362, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:49:19.411477 138027964458816 metric_logger.py:196] completed step: 1, seconds: 1.072, TFLOP/s/device: 12.679, Tokens/s/device: 1911.043, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:49:20.187171 138027964458816 metric_logger.py:196] completed step: 2, seconds: 0.030, TFLOP/s/device: 451.445, Tokens/s/device: 68046.649, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0424 14:49:20.334565 138027964458816 metric_logger.py:196] completed step: 3, seconds: 0.765, TFLOP/s/device: 17.752, Tokens/s/device: 2675.701, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0424 14:49:20.340477 138027964458816 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0424 14:49:20.629404 138027964458816 checkpointing.py:798] Waited 0.28890156745910645 seconds for step 5 to finish before starting checkpointing. I0424 14:49:20.632576 138027964458816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:49:21.961898 2813 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:49:24.031216 137897256257280 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/0/items/array_metadatas/process_3 I0424 14:50:00.654212 137898380752640 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.584 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.02795433998108 s) (per-host) I0424 14:50:00.654318 137898380752640 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.738680s. I0424 14:50:09.128006 137898380752640 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.212351s. I0424 14:50:09.128218 137896732382976 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:50:09.128286 137896732382976 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:50:09.128334 137896732382976 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:50:09.130968 137896732382976 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 14:50:09.131148 138027964458816 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. W0424 14:50:09.131294 138027964458816 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.498718 seconds. If this number is high, consider checkpointing less frequently. I0424 14:50:09.133345 138027964458816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 5 I0424 14:50:09.135835 138027964458816 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/5. I0424 14:50:09.425538 138027964458816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:50:09.425647 138027964458816 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 I0424 14:50:09.460686 138027964458816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.036701s I0424 14:50:09.460860 138027964458816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 34.995 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04408001899719238 s) (per-host) I0424 14:50:09.460912 138027964458816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044142s (batch_requests_ready=0.002750s, total_serialization_initiated=0.041323s, others=0.000070s) I0424 14:50:09.461021 138027964458816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049142s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.049127) I0424 14:50:09.461717 138027964458816 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.33 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/5. I0424 14:50:09.462111 137896732382976 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:10:09.462074 I0424 14:50:09.488123 138027964458816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:50:09.488396 137897256257280 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:50:09.488523 138027964458816 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042160.6325474, 'wait_for_prev_duration_secs': 48.49871802330017, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777042209.1333838, 'checkpointer_blocking_duration_secs': 0.32886672019958496, 'get_old_steps_start_time': 1777042209.4622738, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1777042160.6297429, 'checkpoint_manager_blocking_duration_secs': 48.85874581336975} I0424 14:50:09.488686 138027964458816 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0424 14:50:09.489719 138027964458816 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.572, Tokens/s/device: 12446.216, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0424 14:50:09.497570 138027964458816 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.360, Tokens/s/device: 13921.461, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0424 14:50:14.766411 137874187966208 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/5/items/array_metadatas/process_3 I0424 14:50:33.308351 138027964458816 metric_logger.py:196] completed step: 6, seconds: 49.155, TFLOP/s/device: 0.276, Tokens/s/device: 41.664, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0424 14:50:33.455978 138027964458816 metric_logger.py:196] completed step: 7, seconds: 23.665, TFLOP/s/device: 0.574, Tokens/s/device: 86.540, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0424 14:50:33.603532 138027964458816 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.519, Tokens/s/device: 13342.541, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0424 14:50:33.751651 138027964458816 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 14:50:33.752327 138027964458816 checkpointing.py:798] Waited 0.0006885528564453125 seconds for step 9 to finish before starting checkpointing. I0424 14:50:33.754317 138027964458816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:50:50.757972 137896732382976 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 38.209 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.34115266799927 s) (per-host) I0424 14:50:50.758087 137896732382976 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.295861s. I0424 14:51:00.058804 137896732382976 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.596568s. I0424 14:51:00.059051 137897256257280 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:51:00.059114 137897256257280 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:51:00.059168 137897256257280 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:51:00.060618 137897256257280 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0424 14:51:00.060777 138027964458816 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0424 14:51:00.060902 138027964458816 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.306589 seconds. If this number is high, consider checkpointing less frequently. I0424 14:51:00.062341 138027964458816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0424 14:51:00.064428 138027964458816 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 14:51:00.381597 138027964458816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:51:00.381691 138027964458816 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 I0424 14:51:00.416377 138027964458816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.036409s I0424 14:51:00.416518 138027964458816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 35.524 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04342293739318848 s) (per-host) I0424 14:51:00.416564 138027964458816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.043479s (batch_requests_ready=0.002868s, total_serialization_initiated=0.040550s, others=0.000061s) I0424 14:51:00.416646 138027964458816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.047538s (all_items=0.000010s, per_item={'items': '0.00001025'}, temp_paths=0.047528) I0424 14:51:00.417341 138027964458816 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 14:51:00.417630 137897256257280 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:11:00.417598 I0424 14:51:00.420680 138027964458816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:51:00.420973 137874187966208 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:51:00.421134 138027964458816 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777042233.7542872, 'wait_for_prev_duration_secs': 26.306588888168335, 'time_between_consecutive_saves_sec': 24.623278617858887, 'checkpointer_blocking_start_time': 1777042260.0623817, 'checkpointer_blocking_duration_secs': 0.355391263961792, 'get_old_steps_start_time': 1777042260.417791, 'get_old_steps_duration_secs': 2.5033950805664062e-05, 'checkpoint_manager_blocking_start_time': 1777042233.7525592, 'checkpoint_manager_blocking_duration_secs': 26.66854166984558} I0424 14:51:00.421258 138027964458816 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 14:51:00.421301 138027964458816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:51:05.928695 137894039635712 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9/items/array_metadatas/process_3 I0424 14:51:44.127086 137897256257280 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 36.102 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.75395131111145 s) (per-host) I0424 14:51:44.127209 137897256257280 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.709468s. I0424 14:51:49.305445 137897256257280 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 48.887686s. I0424 14:51:49.305778 137874187966208 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:51:49.305903 137874187966208 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:51:49.305960 137874187966208 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:51:49.307427 137874187966208 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 14:51:49.307657 138027964458816 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. I0424 14:51:49.307820 138027964458816 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:51:49.308719 138027964458816 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.787, Tokens/s/device: 13835.127, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 14:52:02 UTC 2026 EXIT_CODE=0