XPK Start: Mon Apr 20 18:01:04 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 18:01:28.366288: 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) I0420 18:01:28.544224 136865169127232 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 18:01:37.584765 136865169127232 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 18:01:37.587054 136865169127232 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-afkwr-slice-job-0-0.mt-09-pdb-lt-1-afkwr:8482 I0420 18:01:38.895838 136865169127232 max_utils.py:284] Jax distributed system initialized! I0420 18:01:45.307025 136865169127232 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 18:01:45.307126 136865169127232 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 18:01:45.307165 136865169127232 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 I0420 18:01:45.307200 136865169127232 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 18:01:46.006631 136865169127232 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0420 18:01:46.118332 136865169127232 checkpointing.py:688] Setting up checkpoint logger... I0420 18:01:46.118443 136865169127232 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 18:01:46.118491 136865169127232 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 18:01:46.118704 136865169127232 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 0x7c79cc7180b0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 18:01:49.120634 136865169127232 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 18:01:49.120907 136865169127232 checkpoint_manager.py:708] [process=1][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c64bc6612b0>}, handler_registry=None I0420 18:01:49.121142 136865169127232 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c64bc6612b0>` 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`. I0420 18:01:49.121188 136865169127232 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c65a44c2c60>` 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`. I0420 18:01:49.121223 136865169127232 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 0x7c64bc6612b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c64bc6612b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c65a44c2c60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c65a44c2c60>}). I0420 18:01:49.121540 136865169127232 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 18:01:49.121610 136865169127232 async_checkpointer.py:192] [process=1][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c64bc5a8fe0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 18:01:50.027157 136865169127232 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints I0420 18:01:50.432134 136865169127232 checkpoint_manager.py:929] [process=1][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=10), 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_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c64bc6623f0> I0420 18:01:50.432308 136865169127232 checkpointing.py:302] Checkpoint manager created! I0420 18:01:51.394144 136865169127232 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 18:01:51.394266 136865169127232 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0420 18:01:52.832370 136865169127232 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: ('tensor',) I0420 18:01:52.832483 136865169127232 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0420 18:01:52.832534 136865169127232 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0420 18:01:52.832576 136865169127232 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: ('tensor', None, 'fsdp') I0420 18:01:52.832619 136865169127232 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0420 18:01:52.832664 136865169127232 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0420 18:01:52.832704 136865169127232 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0420 18:01:52.832739 136865169127232 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: ('tensor', None, None, 'fsdp') I0420 18:01:52.832774 136865169127232 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0420 18:01:52.832808 136865169127232 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0420 18:01:52.832840 136865169127232 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', 'tensor') I0420 18:01:52.832871 136865169127232 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: ('tensor', 'fsdp') I0420 18:01:53.244425 136865169127232 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:01:53.244517 136865169127232 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 18:01:53.249805 136865169127232 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 18:01:53.249866 136865169127232 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 18:01:53.266225 136865169127232 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:01:53.266284 136865169127232 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 18:01:53.281739 136865169127232 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 18:01:53.281794 136865169127232 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 18:01:53.306545 136865169127232 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:01:53.306612 136865169127232 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 18:01:53.322240 136865169127232 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:01:53.322300 136865169127232 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 18:01:53.337859 136865169127232 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 18:01:53.337919 136865169127232 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 18:01:53.367331 136865169127232 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 18:01:53.367407 136865169127232 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 18:01:53.387284 136865169127232 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 18:01:53.387351 136865169127232 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0420 18:02:08.523390 136865169127232 max_utils.py:791] Total memory size: 0.9 GB, Output size: 0.4 GB, Temp size: 0.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0420 18:02:08.525839 136865169127232 metric_logger.py:301] number parameters: 1.104 billion I0420 18:02:09.462767 136865169127232 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 18:02:24.146400 136865169127232 checkpointing.py:798] Waited 14.683619022369385 seconds for step 0 to finish before starting checkpointing. I0420 18:02:24.149240 136865169127232 checkpoint_manager.py:2009] [process=1][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:02:24.151120 136865169127232 checkpoint_manager.py:1512] [process=1] Saving checkpoint at step 0 I0420 18:02:24.152531 136865169127232 event_tracking.py:70] [process=1] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0. I0420 18:02:24.483181 136865169127232 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 18:02:24.484358 136865169127232 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:02:24.484415 136865169127232 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 I0420 18:02:24.754121 136865169127232 base_pytree_checkpoint_handler.py:154] [process=1][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.271730s I0420 18:02:24.754285 136865169127232 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/blocking_gbytes_per_sec: 5.493 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2808499336242676 s) (per-host) I0420 18:02:24.754336 136865169127232 base_pytree_checkpoint_handler.py:768] [process=1][thread=MainThread] Initiated Pytree async_save. Time taken: 0.280910s (batch_requests_ready=0.002995s, total_serialization_initiated=0.277847s, others=0.000067s) I0420 18:02:24.754430 136865169127232 composite_checkpoint_handler.py:715] [process=1][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.284956s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.284937) I0420 18:02:24.755257 136865169127232 event_tracking.py:125] [process=1] [async] Finished blocking save in 0.60 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0. I0420 18:02:24.755602 136736860534528 async_checkpointer.py:76] [process=1][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:22:24.755565 I0420 18:02:24.765586 136865169127232 checkpoint_manager.py:1560] [process=1][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:02:24.765868 136736843749120 async_checkpointer.py:280] [process=1][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:02:24.766026 136865169127232 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776708144.1492226, 'wait_for_prev_duration_secs': 6.0558319091796875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776708144.1511588, 'checkpointer_blocking_duration_secs': 0.6046104431152344, 'get_old_steps_start_time': 1776708144.7557952, 'get_old_steps_duration_secs': 3.0279159545898438e-05, 'checkpoint_manager_blocking_start_time': 1776708144.1468158, 'checkpoint_manager_blocking_duration_secs': 0.6191704273223877} I0420 18:02:24.766142 136865169127232 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 18:02:24.766214 136865169127232 max_utils.py:750] Memstats: After params initialized: I0420 18:02:24.766266 136865169127232 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_2(process=1,(2,0,0,0)) I0420 18:02:24.766304 136865169127232 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_3(process=1,(3,0,0,0)) I0420 18:02:24.766330 136865169127232 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_6(process=1,(2,1,0,0)) I0420 18:02:24.766356 136865169127232 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_7(process=1,(3,1,0,0)) I0420 18:02:25.079041 136865169127232 metric_logger.py:196] completed step: 0, seconds: 0.936, TFLOP/s/device: 3.630, Tokens/s/device: 547.108, total_weights: 16384, loss: 10.864, lm_loss: 10.864, perplexity: 52258.910 I0420 18:02:25.173647 136865169127232 metric_logger.py:196] completed step: 1, seconds: 15.614, TFLOP/s/device: 0.218, Tokens/s/device: 32.790, total_weights: 16384, loss: 10.864, lm_loss: 10.864, perplexity: 52258.910 I0420 18:02:25.612250 136865169127232 metric_logger.py:196] completed step: 2, seconds: 0.038, TFLOP/s/device: 88.532, Tokens/s/device: 13344.454, total_weights: 16384, loss: 9.760, lm_loss: 9.760, perplexity: 17319.064 I0420 18:02:25.670732 136865169127232 metric_logger.py:196] completed step: 3, seconds: 0.428, TFLOP/s/device: 7.940, Tokens/s/device: 1196.799, total_weights: 16384, loss: 8.820, lm_loss: 8.820, perplexity: 6771.344 I0420 18:02:25.788307 136865169127232 metric_logger.py:196] completed step: 4, seconds: 0.075, TFLOP/s/device: 45.114, Tokens/s/device: 6800.011, total_weights: 16384, loss: 7.947, lm_loss: 7.947, perplexity: 2825.893 I0420 18:02:25.796002 136865169127232 metric_logger.py:196] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.819, Tokens/s/device: 8865.801, total_weights: 16384, loss: 7.209, lm_loss: 7.209, perplexity: 1352.040 I0420 18:02:28.306433 2503 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 18:02:30.300569 136741112723200 array_metadata_store.py:203] [process=1][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_1 I0420 18:02:39.851092 136865169127232 metric_logger.py:196] completed step: 6, seconds: 0.119, TFLOP/s/device: 28.656, Tokens/s/device: 4319.290, total_weights: 16384, loss: 6.651, lm_loss: 6.651, perplexity: 773.347 I0420 18:02:39.909505 136865169127232 metric_logger.py:196] completed step: 7, seconds: 13.998, TFLOP/s/device: 0.243, Tokens/s/device: 36.576, total_weights: 16384, loss: 6.267, lm_loss: 6.267, perplexity: 526.951 I0420 18:02:39.967993 136865169127232 metric_logger.py:196] completed step: 8, seconds: 0.064, TFLOP/s/device: 53.060, Tokens/s/device: 7997.751, total_weights: 16384, loss: 6.030, lm_loss: 6.030, perplexity: 415.600 I0420 18:02:40.027072 136865169127232 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 18:02:40.027729 136865169127232 checkpointing.py:798] Waited 0.0006647109985351562 seconds for step 9 to finish before starting checkpointing. I0420 18:02:40.029642 136865169127232 checkpoint_manager.py:2020] [process=1][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:03:01.277776 136736860534528 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/gbytes_per_sec: 42.919 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.80431532859802 s) (per-host) I0420 18:03:01.277882 136736860534528 async_checkpointer.py:90] [process=1][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.522151s. I0420 18:03:07.840833 136736860534528 async_checkpointer.py:160] [process=1][thread=async_save] Background save thread done. Time taken: 43.085088s. I0420 18:03:07.841181 136736843749120 async_checkpointer.py:288] [process=1][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:03:07.841318 136736843749120 async_checkpointer.py:298] [process=1][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:03:07.841368 136736843749120 checkpoint_manager.py:2137] [process=1][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:03:07.842843 136736843749120 checkpoint_manager.py:2146] [process=1][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 18:03:07.843040 136865169127232 checkpoint_manager.py:2032] [process=1][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 18:03:07.843178 136865169127232 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.813536 seconds. If this number is high, consider checkpointing less frequently. I0420 18:03:07.845064 136865169127232 checkpoint_manager.py:1512] [process=1] Saving checkpoint at step 9 I0420 18:03:07.847165 136865169127232 event_tracking.py:70] [process=1] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9. I0420 18:03:08.136475 136865169127232 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0420 18:03:08.136565 136865169127232 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 I0420 18:03:08.162702 136865169127232 base_pytree_checkpoint_handler.py:154] [process=1][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027792s I0420 18:03:08.162875 136865169127232 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/blocking_gbytes_per_sec: 44.111 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03497052192687988 s) (per-host) I0420 18:03:08.162925 136865169127232 base_pytree_checkpoint_handler.py:768] [process=1][thread=MainThread] Initiated Pytree async_save. Time taken: 0.035030s (batch_requests_ready=0.002661s, total_serialization_initiated=0.032301s, others=0.000068s) I0420 18:03:08.163019 136865169127232 composite_checkpoint_handler.py:715] [process=1][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.039042s (all_items=0.000014s, per_item={'items': '0.00001431'}, temp_paths=0.039027) I0420 18:03:08.163728 136865169127232 event_tracking.py:125] [process=1] [async] Finished blocking save in 0.32 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9. I0420 18:03:08.164066 136736843749120 async_checkpointer.py:76] [process=1][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 18:23:08.164026 I0420 18:03:08.169491 136865169127232 checkpoint_manager.py:1560] [process=1][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 18:03:08.169781 136741112723200 async_checkpointer.py:280] [process=1][thread=save_finalize] Waiting for background save thread=async_save. I0420 18:03:08.169949 136865169127232 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776708160.0296113, 'wait_for_prev_duration_secs': 27.813536405563354, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776708187.8451025, 'checkpointer_blocking_duration_secs': 0.31911230087280273, 'get_old_steps_start_time': 1776708188.1642375, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1776708160.0279725, 'checkpoint_manager_blocking_duration_secs': 28.141942739486694} I0420 18:03:08.170063 136865169127232 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 18:03:08.170107 136865169127232 checkpoint_manager.py:2020] [process=1][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 18:03:13.409847 136734151001856 array_metadata_store.py:203] [process=1][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/nnx_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_1 I0420 18:03:48.934848 136736843749120 base_pytree_checkpoint_handler.py:130] [process=1] /jax/orbax/write/gbytes_per_sec: 38.709 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.80690240859985 s) (per-host) I0420 18:03:48.934963 136736843749120 async_checkpointer.py:90] [process=1][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.770779s. I0420 18:03:56.155319 136736843749120 async_checkpointer.py:160] [process=1][thread=async_save] Background save thread done. Time taken: 47.991122s. I0420 18:03:56.155556 136741112723200 async_checkpointer.py:288] [process=1][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 18:03:56.155622 136741112723200 async_checkpointer.py:298] [process=1][thread=save_finalize] No errors found in background save thread=async_save. I0420 18:03:56.155683 136741112723200 checkpoint_manager.py:2137] [process=1][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 18:03:56.158365 136741112723200 checkpoint_manager.py:2146] [process=1][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 18:03:56.158525 136865169127232 checkpoint_manager.py:2032] [process=1][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 18:03:56.158710 136865169127232 checkpoint_manager.py:2009] [process=1][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 18:03:56.159560 136865169127232 metric_logger.py:196] completed step: 9, seconds: 0.059, TFLOP/s/device: 57.958, Tokens/s/device: 8736.009, total_weights: 16384, loss: 5.890, lm_loss: 5.890, perplexity: 361.390 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 18:04:04 UTC 2026 EXIT_CODE=0