XPK Start: Sun Apr 19 02:30:33 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-19 02:30:57.371187: 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) I0419 02:30:57.548950 139418004363072 max_utils.py:273] Attempting to initialize the jax distributed system... I0419 02:31:06.590240 139418004363072 distributed.py:149] Starting JAX distributed service on [::]:8482 I0419 02:31:06.592525 139418004363072 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-z0mf0-slice-job-0-0.mt-15-ocdbt-false-sv-z0mf0:8482 I0419 02:31:07.972584 139418004363072 max_utils.py:284] Jax distributed system initialized! I0419 02:31:14.231442 139418004363072 max_utils.py:800] System Information: Jax Version: 0.9.2 I0419 02:31:14.231543 139418004363072 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0419 02:31:14.231584 139418004363072 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 I0419 02:31:14.231619 139418004363072 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 02:31:15.279362 139418004363072 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 02:31:15.392268 139418004363072 checkpointing.py:688] Setting up checkpoint logger... I0419 02:31:15.392379 139418004363072 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0419 02:31:15.392428 139418004363072 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0419 02:31:15.392630 139418004363072 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7ecc2ab186e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 02:31:18.276800 139418004363072 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 02:31:18.277045 139418004363072 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eb7dc219fd0>}, handler_registry=None I0419 02:31:18.277286 139418004363072 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eb7dc219fd0>` 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`. I0419 02:31:18.277334 139418004363072 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb7dc2461e0>` 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`. I0419 02:31:18.277370 139418004363072 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 0x7eb7dc219fd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7eb7dc219fd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb7dc2461e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7eb7dc2461e0>}). I0419 02:31:18.277701 139418004363072 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0419 02:31:18.277771 139418004363072 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7eb70c5f44a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0419 02:31:20.726077 139418004363072 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints I0419 02:31:20.728392 139418004363072 checkpoint_manager.py:929] [process=0][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_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7eb7dc0f6030> I0419 02:31:20.728502 139418004363072 checkpointing.py:302] Checkpoint manager created! I0419 02:31:21.670759 139418004363072 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 02:31:21.670883 139418004363072 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 02:31:23.036964 139418004363072 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0419 02:31:23.037080 139418004363072 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:31:23.037132 139418004363072 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0419 02:31:23.037176 139418004363072 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0419 02:31:23.037214 139418004363072 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:31:23.037249 139418004363072 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0419 02:31:23.037285 139418004363072 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:31:23.037321 139418004363072 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0419 02:31:23.037354 139418004363072 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:31:23.037387 139418004363072 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0419 02:31:23.037418 139418004363072 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0419 02:31:23.037456 139418004363072 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0419 02:31:23.438518 139418004363072 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:31:23.438611 139418004363072 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:31:23.443974 139418004363072 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 02:31:23.444028 139418004363072 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:31:23.460495 139418004363072 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:31:23.460553 139418004363072 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:31:23.476500 139418004363072 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 02:31:23.476559 139418004363072 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 02:31:23.500587 139418004363072 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:31:23.500655 139418004363072 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:31:23.516598 139418004363072 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:31:23.516661 139418004363072 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:31:23.532557 139418004363072 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 02:31:23.532616 139418004363072 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:31:23.563046 139418004363072 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 02:31:23.563117 139418004363072 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 02:31:23.583311 139418004363072 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 02:31:23.583375 139418004363072 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 02:31:35.059386 139418004363072 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. I0419 02:31:35.069796 139418004363072 metric_logger.py:301] number parameters: 1.104 billion I0419 02:31:46.760189 139418004363072 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 02:31:46.912712 139418004363072 checkpointing.py:798] Waited 0.15250825881958008 seconds for step 0 to finish before starting checkpointing. I0419 02:31:46.915764 139418004363072 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:31:46.917676 139418004363072 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0419 02:31:46.919034 139418004363072 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0. I0419 02:31:47.631065 139418004363072 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 02:31:47.650509 139418004363072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:31:47.650579 139418004363072 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 I0419 02:31:47.912730 139289446049536 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0 I0419 02:31:47.927641 139418004363072 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278567s I0419 02:31:47.928516 139418004363072 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.327 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2895848751068115 s) (per-host) I0419 02:31:47.928590 139418004363072 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.289680s (batch_requests_ready=0.003254s, total_serialization_initiated=0.285630s, others=0.000796s) I0419 02:31:47.928676 139418004363072 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.293639s (all_items=0.000031s, per_item={'items': '0.00003076'}, temp_paths=0.293609) I0419 02:31:47.929418 139418004363072 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0. I0419 02:31:47.929719 139289454442240 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:51:47.929686 I0419 02:31:47.932164 139418004363072 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:31:47.932443 139288351651584 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:31:47.932580 139418004363072 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565906.9157457, 'wait_for_prev_duration_secs': 7.796287536621094e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776565906.9177153, 'checkpointer_blocking_duration_secs': 1.012115716934204, 'get_old_steps_start_time': 1776565907.9298563, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776565906.9131024, 'checkpoint_manager_blocking_duration_secs': 1.0194408893585205} I0419 02:31:47.932684 139418004363072 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 02:31:47.932753 139418004363072 max_utils.py:750] Memstats: After params initialized: I0419 02:31:47.932802 139418004363072 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0419 02:31:47.932852 139418004363072 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0419 02:31:47.932880 139418004363072 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0419 02:31:47.932905 139418004363072 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0419 02:31:48.282094 139418004363072 metric_logger.py:196] completed step: 0, seconds: 11.368, TFLOP/s/device: 1.195, Tokens/s/device: 180.149, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:31:48.286219 139418004363072 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/tensorboard/' I0419 02:31:48.683372 139418004363072 metric_logger.py:196] completed step: 1, seconds: 1.518, TFLOP/s/device: 8.950, Tokens/s/device: 1348.977, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0419 02:31:48.825219 139418004363072 metric_logger.py:196] completed step: 2, seconds: 0.402, TFLOP/s/device: 33.780, Tokens/s/device: 5091.614, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0419 02:31:48.972685 139418004363072 metric_logger.py:196] completed step: 3, seconds: 0.014, TFLOP/s/device: 1000.451, Tokens/s/device: 150798.910, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0419 02:31:48.985936 139418004363072 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0419 02:31:49.267778 139418004363072 checkpointing.py:798] Waited 0.2818138599395752 seconds for step 5 to finish before starting checkpointing. I0419 02:31:49.270675 139418004363072 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:31:50.532875 139289429264128 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/items I0419 02:31:50.701084 139286796039936 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565909842680444, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776565909842680444, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0419 02:31:52.944438 2807 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 02:31:55.160425 139288405006080 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_0 I0419 02:32:03.946681 139288396613376 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 11.020205s (commit=9.981720s, array_metadata_write=1.038485s) I0419 02:32:03.948142 139289454442240 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 96.854 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 16.309163093566895 s) (per-host) I0419 02:32:03.948316 139289454442240 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 16.017967s. I0419 02:32:05.010562 139289454442240 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565909842680444, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0419 02:32:05.706434 139286796039936 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776565909842680444, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA W0419 02:32:05.765773 139289454442240 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/items, I0419 02:32:06.643404 139289454442240 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000770s. I0419 02:32:06.643683 139289454442240 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.507662s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/items I0419 02:32:06.645570 139289454442240 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0/items I0419 02:32:07.164210 139289454442240 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0 I0419 02:32:08.567375 139289454442240 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0`. I0419 02:32:08.568403 139289454442240 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 21.65 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/0 I0419 02:32:08.569724 139289454442240 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 20.639372s. I0419 02:32:08.570010 139288351651584 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:32:08.570172 139288351651584 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:32:08.570281 139288351651584 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:32:08.573089 139288351651584 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 02:32:08.573311 139418004363072 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 02:32:08.573418 139418004363072 checkpoint_manager.py:1452] Waiting for previous save to complete took 19.302763 seconds. If this number is high, consider checkpointing less frequently. I0419 02:32:08.575429 139418004363072 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 5 I0419 02:32:08.577738 139418004363072 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5. I0419 02:32:09.301502 139418004363072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:32:09.301634 139418004363072 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 I0419 02:32:09.344442 139418004363072 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045872s I0419 02:32:09.350799 139418004363072 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 21.812 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.07072043418884277 s) (per-host) I0419 02:32:09.350943 139418004363072 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.070894s (batch_requests_ready=0.006660s, total_serialization_initiated=0.057936s, others=0.006298s) I0419 02:32:09.351087 139418004363072 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.075810s (all_items=0.000025s, per_item={'items': '0.00002456'}, temp_paths=0.075785) I0419 02:32:09.352038 139418004363072 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.78 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5. I0419 02:32:09.352428 139288351651584 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:52:09.352390 I0419 02:32:09.354870 139418004363072 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:32:09.355185 139287852996352 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:32:09.355369 139418004363072 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565909.2706306, 'wait_for_prev_duration_secs': 19.30276322364807, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776565928.5754962, 'checkpointer_blocking_duration_secs': 0.7771236896514893, 'get_old_steps_start_time': 1776565929.352649, 'get_old_steps_duration_secs': 4.100799560546875e-05, 'checkpoint_manager_blocking_start_time': 1776565909.2683077, 'checkpoint_manager_blocking_duration_secs': 20.087008714675903} I0419 02:32:09.355519 139418004363072 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0419 02:32:09.510779 139289454442240 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5 I0419 02:32:12.079665 139288935130880 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5/items I0419 02:32:16.196489 139288397879040 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_0 I0419 02:32:24.320528 139287869781760 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.290365s (commit=9.171103s, array_metadata_write=1.119262s) I0419 02:32:24.321686 139288351651584 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 105.015 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 15.041597604751587 s) (per-host) I0419 02:32:24.321823 139288351651584 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 14.969230s. W0419 02:32:26.323755 139288351651584 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5/items, I0419 02:32:27.166490 139288351651584 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000408s. I0419 02:32:27.166707 139288351651584 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.527826s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5/items I0419 02:32:27.168543 139288351651584 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5/items I0419 02:32:28.105636 139288351651584 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5 I0419 02:32:29.490589 139288351651584 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5`. I0419 02:32:29.491591 139288351651584 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 20.92 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/5 I0419 02:32:29.492859 139288351651584 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 20.140276s. I0419 02:32:29.493064 139287852996352 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:32:29.493205 139287852996352 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:32:29.493278 139287852996352 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:32:29.495524 139287852996352 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0419 02:32:32.314988 139418004363072 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 94.841, Tokens/s/device: 14295.487, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0419 02:32:32.326299 139418004363072 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.514, Tokens/s/device: 13944.684, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0419 02:32:32.470577 139418004363072 metric_logger.py:196] completed step: 6, seconds: 43.340, TFLOP/s/device: 0.314, Tokens/s/device: 47.255, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0419 02:32:32.618015 139418004363072 metric_logger.py:196] completed step: 7, seconds: 0.009, TFLOP/s/device: 1500.511, Tokens/s/device: 226173.385, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0419 02:32:32.765860 139418004363072 metric_logger.py:196] completed step: 8, seconds: 0.146, TFLOP/s/device: 93.220, Tokens/s/device: 14051.072, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0419 02:32:32.914237 139418004363072 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 02:32:32.914929 139418004363072 checkpointing.py:798] Waited 0.0007035732269287109 seconds for step 9 to finish before starting checkpointing. I0419 02:32:32.916858 139418004363072 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:32:32.918370 139418004363072 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0419 02:32:32.919860 139418004363072 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9. I0419 02:32:33.627962 139418004363072 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0419 02:32:33.628059 139418004363072 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 I0419 02:32:33.655628 139418004363072 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.029134s I0419 02:32:33.656445 139418004363072 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 40.916 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03770112991333008 s) (per-host) I0419 02:32:33.656521 139418004363072 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037803s (batch_requests_ready=0.002747s, total_serialization_initiated=0.034321s, others=0.000736s) I0419 02:32:33.656609 139418004363072 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041690s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.041673) I0419 02:32:33.657301 139418004363072 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9. I0419 02:32:33.657575 139287852996352 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-19 02:52:33.657550 I0419 02:32:33.663179 139418004363072 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 02:32:33.663408 139287836210944 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0419 02:32:33.663530 139418004363072 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776565952.9168246, 'wait_for_prev_duration_secs': 7.128715515136719e-05, 'time_between_consecutive_saves_sec': 3.421261787414551, 'checkpointer_blocking_start_time': 1776565952.9184077, 'checkpointer_blocking_duration_secs': 0.7392652034759521, 'get_old_steps_start_time': 1776565953.6576917, 'get_old_steps_duration_secs': 2.6464462280273438e-05, 'checkpoint_manager_blocking_start_time': 1776565952.915162, 'checkpoint_manager_blocking_duration_secs': 0.7483346462249756} I0419 02:32:33.663640 139418004363072 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 02:32:33.663687 139418004363072 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 02:32:34.287612 139288351651584 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9 I0419 02:32:36.445583 139288935130880 atomicity.py:140] Creating tmp directory gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/items I0419 02:32:40.121598 139287867537152 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_0 I0419 02:32:48.125696 139287844603648 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 10.592204s (commit=9.132634s, array_metadata_write=1.459570s) I0419 02:32:48.126725 139287852996352 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 108.878 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 14.50798487663269 s) (per-host) I0419 02:32:48.126785 139287852996352 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 14.469134s. W0419 02:32:49.595871 139287852996352 ocdbt_utils.py:168] [process=0][thread=async_save] Skipping merge of OCDBT checkpoints: No per-process OCDBT checkpoint subdirs found in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/items, I0419 02:32:50.029622 139287852996352 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000452s. I0419 02:32:50.029865 139287852996352 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.077774s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/items I0419 02:32:50.031234 139287852996352 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9/items I0419 02:32:50.543354 139287852996352 atomicity.py:666] Finalizing gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9 I0419 02:32:51.887956 139287852996352 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9`. I0419 02:32:51.888799 139287852996352 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 18.97 seconds @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260418_235007/nnx_xpk_feat_nnx_post_train_fixes_20260418_235007_15_ocdbt_false/checkpoints/9 I0419 02:32:51.890124 139287852996352 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 18.232468s. I0419 02:32:51.890257 139287836210944 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 02:32:51.890319 139287836210944 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0419 02:32:51.890367 139287836210944 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 02:32:51.892602 139287836210944 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 02:32:51.892766 139418004363072 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 02:32:51.892955 139418004363072 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 02:32:51.893957 139418004363072 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.313, Tokens/s/device: 13914.461, 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: Sun Apr 19 02:33:03 UTC 2026 EXIT_CODE=0