XPK Start: Tue Apr 21 11:52:16 UTC 2026 2026-04-21 11:53:14.771452: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0421 11:53:14.988308 133090177021760 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-21 11:53:24,030:jax._src.distributed:149: Starting JAX distributed service on [::]:8482 I0421 11:53:24.030872 133090177021760 distributed.py:149] Starting JAX distributed service on [::]:8482 INFO:2026-04-21 11:53:24,033:jax._src.distributed:166: Connecting to JAX distributed service on mt-02-sft-linen-ckpt-ijfsg-slice-job-0-0.mt-02-sft-linen-ckpt-ijfsg:8482 I0421 11:53:24.033110 133090177021760 distributed.py:166] Connecting to JAX distributed service on mt-02-sft-linen-ckpt-ijfsg-slice-job-0-0.mt-02-sft-linen-ckpt-ijfsg:8482 I0421 11:53:46.230676 133090177021760 max_utils.py:284] Jax distributed system initialized! I0421 11:53:51.298908 133090177021760 max_utils.py:800] System Information: Jax Version: 0.8.3 I0421 11:53:51.299012 133090177021760 max_utils.py:801] System Information: Jaxlib Version: 0.8.3 I0421 11:53:51.299052 133090177021760 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Dec 15 2025 14:03:46 (1765836226) cl/844590465 I0421 11:53:51.302496 133090177021760 maxtext_utils.py:1718] Num_devices: 32, shape (1, 4, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 11:53:51.497492 133090177021760 maxtext_utils.py:1718] Num_devices: 32, shape (1, 4, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 11:53:52.585802 133090177021760 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 11:53:52.586246 133090177021760 base_pytree_checkpoint_handler.py:411] 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 0x790ac2f945c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 11:53:52.586307 133090177021760 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 W0421 11:53:53.093638 133090177021760 checkpoint.py:202] Metadata file does not exist: gs://lance-maxtext/pt_seed_ckpts/pt_seed_ckpts/pt_seed_ckpt_gpt352k_linen/checkpoints/9/items/_CHECKPOINT_METADATA I0421 11:53:54.046857 1902 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 11:53:55.184420 133090177021760 checkpointer.py:304] Restoring checkpoint from gs://lance-maxtext/pt_seed_ckpts/pt_seed_ckpts/pt_seed_ckpt_gpt352k_linen/checkpoints/9/items. W0421 11:53:57.023326 133090177021760 transform_utils.py:230] The transformations API will eventually be replaced by an upgraded design. The current API will not be removed until this point, but it will no longer be actively worked on. I0421 11:53:57.023705 133090177021760 transform_utils.py:288] The following keys are not loaded from the original tree after applying specified transforms: params/params/decoder/to_nnx__rngs/aqt/count, params/params/decoder/to_nnx__rngs/aqt/key, params/params/decoder/to_nnx__rngs/dropout/count, params/params/decoder/to_nnx__rngs/dropout/key, params/params/decoder/to_nnx__rngs/params/count, params/params/decoder/to_nnx__rngs/params/key I0421 11:53:57.550569 133090177021760 checkpointer.py:318] Finished restoring checkpoint in 2.74 seconds from gs://lance-maxtext/pt_seed_ckpts/pt_seed_ckpts/pt_seed_ckpt_gpt352k_linen/checkpoints/9/items. I0421 11:53:57.613933 133090177021760 config.py:112] TensorFlow version 2.20.0 available. I0421 11:53:57.614433 133090177021760 config.py:125] JAX version 0.8.3 available. /deps/src/maxtext/input_pipeline/input_pipeline_utils.py:467: UserWarning: WARNING: Inefficient dataloading. Your train or eval dataset contains 3 shards, smaller than number of host loading data. This is known to lead to inefficient dataloading. Seegithub.com/google/maxtext/blob/main/getting_started/Data_Input_Pipeline.md#multihost-dataloading-best-practice warnings.warn( E0421 11:54:03.422085 133090177021760 packing.py:209] PackAndBatchOperation is deprecated. Please use lazy_dataset.FirstFitPackIterDataset instead. I0421 11:54:03.422305 133090177021760 data_loader.py:408] Adding CopyNumPyArrayToSharedMemory MapTransform. I0421 11:54:03.802217 133090177021760 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 11:54:03.802365 133090177021760 base_pytree_checkpoint_handler.py:411] 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 0x790ac2f945c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 11:54:03.802412 133090177021760 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0421 11:54:03.802448 133090177021760 base_pytree_checkpoint_handler.py:411] 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 0x790ac2f945c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 11:54:03.802492 133090177021760 checkpoint_manager.py:702] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=None, item_handlers={'model_params': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f1e04658b0>, 'optimizer_state': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f3d0cf1cd0>, 'custom_metadata': <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0ceaa50>}, handler_registry=None I0421 11:54:03.802697 133090177021760 composite_checkpoint_handler.py:237] Deferred registration for item: "model_params". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f1e04658b0>` for item "model_params" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0421 11:54:03.802754 133090177021760 composite_checkpoint_handler.py:237] Deferred registration for item: "optimizer_state". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f3d0cf1cd0>` for item "optimizer_state" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`. I0421 11:54:03.802786 133090177021760 composite_checkpoint_handler.py:237] Deferred registration for item: "custom_metadata". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0ceaa50>` for item "custom_metadata" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0421 11:54:03.802813 133090177021760 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0cf0590>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`. I0421 11:54:03.802842 133090177021760 composite_checkpoint_handler.py:505] Initialized registry DefaultCheckpointHandlerRegistry({('model_params', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f1e04658b0>, ('model_params', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f1e04658b0>, ('optimizer_state', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f3d0cf1cd0>, ('optimizer_state', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78f3d0cf1cd0>, ('custom_metadata', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0ceaa50>, ('custom_metadata', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0ceaa50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0cf0590>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78f3d0cf0590>}). I0421 11:54:03.803303 133090177021760 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78f3d0d41e40> timeout: 600 secs and primary_host=0 for async checkpoint writes I0421 11:54:06.522577 133090177021760 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://lance-maxtext/pt_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114106/pt_sft_linen_xpk_feat_nnx_post_train_fixes_20260421_114106_02_sft_linen_ckpt/checkpoints I0421 11:54:06.524700 133090177021760 checkpoint_manager.py:921] [process=7][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=10000, 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_hns=False, 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=None, preservation_policy=None, prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://lance-maxtext/pt_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114106/pt_sft_linen_xpk_feat_nnx_post_train_fixes_20260421_114106_02_sft_linen_ckpt/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78f3d0cf18b0> I0421 11:54:06.524968 133090177021760 peft_trainer.py:584] Training with mesh: Mesh('diloco': 1, 'data': 4, 'stage': 1, 'fsdp': 8, 'fsdp_transpose': 1, 'sequence': 1, 'context': 1, 'context_autoregressive': 1, 'tensor': 1, 'tensor_transpose': 1, 'tensor_sequence': 1, 'expert': 1, 'autoregressive': 1, axis_types=(Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto, Auto)) I0421 11:54:06.940661 133090177021760 peft_trainer.py:594] Compiled train_step cache size: 0 I0421 11:54:06.942645 133090177021760 metric_logger.py:301] number parameters: 0.000 billion I0421 11:54:06.945188 132940134590208 grain_pool.py:367] Grain pool will use 1 processes. I0421 11:54:06.972497 132940134590208 grain_pool.py:440] Grain pool will start child processes. Per train step: Total TFLOPs: 0.00 split as 54.29% learnable weight flops and 45.71% attention flops I0421 11:54:06.977636 132940134590208 grain_pool.py:448] Grain pool started all child processes. 2026-04-21 11:54:10.977290: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. 2026-04-21 11:54:11.022343: I tensorflow/core/platform/cpu_feature_guard.cc:210] This TensorFlow binary is optimized to use available CPU instructions in performance-critical operations. To enable the following instructions: AVX2 AVX512F AVX512_VNNI AVX512_BF16 FMA, in other operations, rebuild TensorFlow with the appropriate compiler flags. 2026-04-21 11:54:12.190697: I tensorflow/core/util/port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. 2026-04-21 11:54:16.065467: 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) Traceback (most recent call last): File "<frozen runpy>", line 198, in _run_module_as_main File "<frozen runpy>", line 88, in _run_code File "/deps/src/maxtext/trainers/post_train/sft/train_sft.py", line 281, in <module> app.run(main) File "/usr/local/lib/python3.12/site-packages/absl/app.py", line 316, in run _run_main(main, args) File "/usr/local/lib/python3.12/site-packages/absl/app.py", line 261, in _run_main sys.exit(main(argv)) ^^^^^^^^^^ File "/deps/src/maxtext/trainers/post_train/sft/train_sft.py", line 277, in main train(mt_config, goodput_recorder) File "/deps/src/maxtext/trainers/post_train/sft/train_sft.py", line 254, in train trainer = train_model(mt_config, trainer, mesh) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/deps/src/maxtext/trainers/post_train/sft/train_sft.py", line 240, in train_model trainer.train(trainer.data_hooks.train_data_iterator, trainer.data_hooks.eval_data_iterator) File "/usr/local/lib/python3.12/site-packages/tunix/sft/peft_trainer.py", line 692, in train train_loss, aux, grad_norm = train_step(train_example) ^^^^^^^^^^^^^^^^^^^^^^^^^^ ValueError: not enough values to unpack (expected 3, got 2) I0421 11:54:25.100547 132940134590208 grain_pool.py:542] Grain pool is exiting. I0421 11:54:25.100656 132940134590208 grain_pool.py:547] Shutting down multiprocessing system. I0421 11:54:28.204804 132940134590208 grain_pool.py:547] Shutting down multiprocessing system. /usr/local/lib/python3.12/multiprocessing/resource_tracker.py:279: UserWarning: resource_tracker: There appear to be 20 leaked shared_memory objects to clean up at shutdown warnings.warn('resource_tracker: There appear to be %d ' XPK End: Tue Apr 21 11:54:41 UTC 2026 EXIT_CODE=1