XPK Start: Fri Apr 24 14:38:17 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 14:38:42.029247: 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 14:38:42.595811 132636085835584 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:38:51.636378 132636085835584 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:38:51.638751 132636085835584 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-gvlqk-slice-job-0-0.mt-16-zarr3-true-sv-gvlqk:8482 I0424 14:38:52.849461 132636085835584 max_utils.py:284] Jax distributed system initialized! I0424 14:38:58.873728 132636085835584 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:38:58.873832 132636085835584 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:38:58.873872 132636085835584 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 14:38:58.873906 132636085835584 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:38:59.566019 132636085835584 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:38:59.679811 132636085835584 checkpointing.py:688] Setting up checkpoint logger... I0424 14:38:59.679930 132636085835584 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 14:38:59.679975 132636085835584 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:38:59.680243 132636085835584 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 0x78a1329b17c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:39:02.625777 132636085835584 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:39:02.626019 132636085835584 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 0x788cbc5362d0>}, handler_registry=None I0424 14:39:02.626275 132636085835584 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x788cbc5362d0>` 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 14:39:02.626326 132636085835584 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x788cbc16f620>` 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 14:39:02.626361 132636085835584 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 0x788cbc5362d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x788cbc5362d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x788cbc16f620>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x788cbc16f620>}). I0424 14:39:02.626688 132636085835584 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:39:02.626759 132636085835584 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x788cbc1d71a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:39:03.339647 132636085835584 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_true/checkpoints I0424 14:39:03.368474 132636085835584 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=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_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x788cbc537d10> I0424 14:39:03.368612 132636085835584 checkpointing.py:302] Checkpoint manager created! I0424 14:39:04.333863 132636085835584 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:39:04.333984 132636085835584 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 14:39:05.728409 132636085835584 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:39:05.728551 132636085835584 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:39:05.728623 132636085835584 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:39:05.728685 132636085835584 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:39:05.728742 132636085835584 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:39:05.728802 132636085835584 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:39:05.728869 132636085835584 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:39:05.728936 132636085835584 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:39:05.728999 132636085835584 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:39:05.729061 132636085835584 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:39:05.729148 132636085835584 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:39:05.729217 132636085835584 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:39:06.133061 132636085835584 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:39:06.133165 132636085835584 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:39:06.138647 132636085835584 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:39:06.138705 132636085835584 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:39:06.155411 132636085835584 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:39:06.155469 132636085835584 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:39:06.171433 132636085835584 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:39:06.171493 132636085835584 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:39:06.196712 132636085835584 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 14:39:06.196783 132636085835584 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:39:06.212924 132636085835584 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 14:39:06.212995 132636085835584 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:39:06.228938 132636085835584 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 14:39:06.228999 132636085835584 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:39:06.259329 132636085835584 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:39:06.259401 132636085835584 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:39:06.279650 132636085835584 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:39:06.279714 132636085835584 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:39:17.473485 132636085835584 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 14:39:17.475893 132636085835584 metric_logger.py:301] number parameters: 1.104 billion I0424 14:39:28.770911 132636085835584 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 14:39:28.923928 132636085835584 checkpointing.py:798] Waited 0.15300250053405762 seconds for step 0 to finish before starting checkpointing. I0424 14:39:28.926243 132636085835584 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:39:28.927845 132636085835584 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0424 14:39:28.929262 132636085835584 event_tracking.py:70] [process=6] [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_true/checkpoints/0. I0424 14:39:29.225568 132636085835584 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:39:29.228815 132636085835584 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:39:29.228876 132636085835584 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 14:39:29.510551 132636085835584 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.285750s I0424 14:39:29.510728 132636085835584 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.230 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2949199676513672 s) (per-host) I0424 14:39:29.510780 132636085835584 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.294982s (batch_requests_ready=0.003188s, total_serialization_initiated=0.291724s, others=0.000070s) I0424 14:39:29.510879 132636085835584 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.299562s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.299545) I0424 14:39:29.511613 132636085835584 event_tracking.py:125] [process=6] [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_true/checkpoints/0. I0424 14:39:29.511965 132507551340288 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:59:29.511925 I0424 14:39:29.514043 132636085835584 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:39:29.514284 132505399183104 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:39:29.514396 132636085835584 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041568.9262254, 'wait_for_prev_duration_secs': 5.841255187988281e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041568.927883, 'checkpointer_blocking_duration_secs': 0.5842494964599609, 'get_old_steps_start_time': 1777041569.512155, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1777041568.9243436, 'checkpoint_manager_blocking_duration_secs': 0.5900132656097412} I0424 14:39:29.514503 132636085835584 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 14:39:29.514572 132636085835584 max_utils.py:750] Memstats: After params initialized: I0424 14:39:29.514622 132636085835584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 14:39:29.514654 132636085835584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 14:39:29.514680 132636085835584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 14:39:29.514704 132636085835584 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 14:39:29.835965 132636085835584 metric_logger.py:196] completed step: 0, seconds: 11.294, TFLOP/s/device: 1.203, Tokens/s/device: 181.336, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:39:30.022777 132636085835584 metric_logger.py:196] completed step: 1, seconds: 1.063, TFLOP/s/device: 12.781, Tokens/s/device: 1926.485, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:39:30.459324 132636085835584 metric_logger.py:196] completed step: 2, seconds: 0.041, TFLOP/s/device: 331.766, Tokens/s/device: 50007.325, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0424 14:39:30.606891 132636085835584 metric_logger.py:196] completed step: 3, seconds: 0.426, TFLOP/s/device: 31.871, Tokens/s/device: 4803.971, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0424 14:39:30.613035 132636085835584 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0424 14:39:30.901776 132636085835584 checkpointing.py:798] Waited 0.2887125015258789 seconds for step 5 to finish before starting checkpointing. I0424 14:39:30.904602 132636085835584 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:39:32.527736 2789 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:39:34.593652 132505445332736 array_metadata_store.py:203] [process=6][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_true/checkpoints/0/items/array_metadatas/process_6 I0424 14:40:06.576074 132507551340288 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 42.280 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.36022686958313 s) (per-host) I0424 14:40:06.576213 132507551340288 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.064134s. I0424 14:40:13.773890 132507551340288 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 44.261795s. I0424 14:40:13.774119 132505399183104 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:40:13.774190 132505399183104 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:40:13.774240 132505399183104 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:40:13.776086 132505399183104 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 14:40:13.776267 132636085835584 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0424 14:40:13.776407 132636085835584 checkpoint_manager.py:1452] Waiting for previous save to complete took 42.871801 seconds. If this number is high, consider checkpointing less frequently. I0424 14:40:13.778581 132636085835584 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0424 14:40:13.781021 132636085835584 event_tracking.py:70] [process=6] [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_true/checkpoints/5. I0424 14:40:14.398516 132636085835584 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:40:14.398609 132636085835584 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 14:40:14.436217 132636085835584 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.039314s I0424 14:40:14.436391 132636085835584 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 32.965 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04679369926452637 s) (per-host) I0424 14:40:14.436443 132636085835584 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.046855s (batch_requests_ready=0.002747s, total_serialization_initiated=0.044039s, others=0.000069s) I0424 14:40:14.436536 132636085835584 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051622s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.051606) I0424 14:40:14.437281 132636085835584 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.66 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_true/checkpoints/5. I0424 14:40:14.437646 132505399183104 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:00:14.437608 I0424 14:40:14.453764 132636085835584 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:40:14.454075 132482376636160 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:40:14.454261 132636085835584 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041570.9045734, 'wait_for_prev_duration_secs': 42.871800661087036, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041613.7786207, 'checkpointer_blocking_duration_secs': 0.6591312885284424, 'get_old_steps_start_time': 1777041614.437775, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1777041570.9021301, 'checkpoint_manager_blocking_duration_secs': 43.55209565162659} I0424 14:40:14.454376 132636085835584 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0424 14:40:14.455427 132636085835584 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.702, Tokens/s/device: 12465.686, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0424 14:40:14.463231 132636085835584 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.149, Tokens/s/device: 13889.642, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0424 14:40:20.122303 132505445332736 array_metadata_store.py:203] [process=6][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_true/checkpoints/5/items/array_metadatas/process_6 I0424 14:40:37.353274 132636085835584 metric_logger.py:196] completed step: 6, seconds: 43.848, TFLOP/s/device: 0.310, Tokens/s/device: 46.707, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0424 14:40:37.500815 132636085835584 metric_logger.py:196] completed step: 7, seconds: 22.745, TFLOP/s/device: 0.597, Tokens/s/device: 90.042, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0424 14:40:37.648581 132636085835584 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.625, Tokens/s/device: 13358.467, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0424 14:40:37.796569 132636085835584 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 14:40:37.797264 132636085835584 checkpointing.py:798] Waited 0.0007131099700927734 seconds for step 9 to finish before starting checkpointing. I0424 14:40:37.799655 132636085835584 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:40:56.261741 132505399183104 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.724 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.872106075286865 s) (per-host) I0424 14:40:56.261864 132505399183104 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.823565s. I0424 14:41:04.636265 132505399183104 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.197950s. I0424 14:41:04.636558 132482376636160 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:41:04.636683 132482376636160 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:41:04.636731 132482376636160 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:41:04.639042 132482376636160 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0424 14:41:04.639225 132636085835584 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0424 14:41:04.639354 132636085835584 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.839702 seconds. If this number is high, consider checkpointing less frequently. I0424 14:41:04.640772 132636085835584 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0424 14:41:04.642914 132636085835584 event_tracking.py:70] [process=6] [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_true/checkpoints/9. I0424 14:41:04.953222 132636085835584 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:41:04.953321 132636085835584 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 14:41:04.987546 132636085835584 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035864s I0424 14:41:04.987691 132636085835584 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 35.977 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0428767204284668 s) (per-host) I0424 14:41:04.987738 132636085835584 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.042935s (batch_requests_ready=0.002861s, total_serialization_initiated=0.040011s, others=0.000063s) I0424 14:41:04.987825 132636085835584 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.047051s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.047040) I0424 14:41:04.988521 132636085835584 event_tracking.py:125] [process=6] [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_true/checkpoints/9. I0424 14:41:04.988850 132511837427456 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 15:01:04.988810 I0424 14:41:04.999082 132636085835584 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:41:04.999345 132482376636160 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:41:04.999466 132636085835584 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041637.7996275, 'wait_for_prev_duration_secs': 26.839701890945435, 'time_between_consecutive_saves_sec': 24.023492336273193, 'checkpointer_blocking_start_time': 1777041664.640811, 'checkpointer_blocking_duration_secs': 0.34818339347839355, 'get_old_steps_start_time': 1777041664.989013, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1777041637.7974951, 'checkpoint_manager_blocking_duration_secs': 27.20193886756897} I0424 14:41:04.999580 132636085835584 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 14:41:04.999623 132636085835584 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:41:10.393825 132505445332736 array_metadata_store.py:203] [process=6][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_true/checkpoints/9/items/array_metadatas/process_6 I0424 14:41:45.566275 132511837427456 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.886 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.62143588066101 s) (per-host) I0424 14:41:45.566377 132511837427456 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.577414s. I0424 14:41:53.573946 132511837427456 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.584967s. I0424 14:41:53.574187 132482376636160 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:41:53.574253 132482376636160 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:41:53.574294 132482376636160 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:41:53.575607 132482376636160 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 14:41:53.575767 132636085835584 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0424 14:41:53.575944 132636085835584 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:41:53.576982 132636085835584 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.929, Tokens/s/device: 13856.469, 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 14:42:02 UTC 2026 EXIT_CODE=0