XPK Start: Fri Apr 24 13:59:25 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 13:59:50.198115: 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 13:59:50.411208 134612197783360 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 13:59:59.453994 134612197783360 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 13:59:59.456418 134612197783360 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-hy20f-slice-job-0-0.mt-09-pdb-lt-1-hy20f:8482 I0424 14:00:00.650610 134612197783360 max_utils.py:284] Jax distributed system initialized! I0424 14:00:06.953886 134612197783360 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:00:06.954008 134612197783360 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:00:06.954049 134612197783360 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 14:00:06.954085 134612197783360 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:00:07.644983 134612197783360 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0424 14:00:07.758065 134612197783360 checkpointing.py:688] Setting up checkpoint logger... I0424 14:00:07.758182 134612197783360 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 14:00:07.758226 134612197783360 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:00:07.758424 134612197783360 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 0x7a6d3eb18320>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:00:10.728205 134612197783360 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:00:10.728443 134612197783360 checkpoint_manager.py:708] [process=3][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a58d075d4c0>}, handler_registry=None I0424 14:00:10.728688 134612197783360 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a58d075d4c0>` 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 14:00:10.728737 134612197783360 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a58cbe91c40>` 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 14:00:10.728773 134612197783360 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 0x7a58d075d4c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a58d075d4c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a58cbe91c40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a58cbe91c40>}). I0424 14:00:10.729124 134612197783360 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:00:10.729197 134612197783360 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a58cbe7f6a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:00:13.009806 134612197783360 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_09_pdb_lt_1/checkpoints I0424 14:00:13.026637 134612197783360 checkpoint_manager.py:929] [process=3][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=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_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a58b8123b30> I0424 14:00:13.026773 134612197783360 checkpointing.py:302] Checkpoint manager created! I0424 14:00:14.012339 134612197783360 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:00:14.012459 134612197783360 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0424 14:00:15.404940 134612197783360 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: ('tensor',) I0424 14:00:15.405058 134612197783360 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0424 14:00:15.405112 134612197783360 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, 'tensor') I0424 14:00:15.405156 134612197783360 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: ('tensor', None, 'fsdp') I0424 14:00:15.405194 134612197783360 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0424 14:00:15.405230 134612197783360 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: ('tensor', None) I0424 14:00:15.405266 134612197783360 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 14:00:15.405302 134612197783360 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: ('tensor', None, None, 'fsdp') I0424 14:00:15.405338 134612197783360 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 14:00:15.405372 134612197783360 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, 'tensor', None) I0424 14:00:15.405405 134612197783360 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', 'tensor') I0424 14:00:15.405437 134612197783360 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: ('tensor', 'fsdp') I0424 14:00:15.811563 134612197783360 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:00:15.811668 134612197783360 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 14:00:15.817026 134612197783360 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:00:15.817082 134612197783360 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 14:00:15.833542 134612197783360 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:00:15.833603 134612197783360 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 14:00:15.849630 134612197783360 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:00:15.849689 134612197783360 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0424 14:00:15.875071 134612197783360 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:00:15.875141 134612197783360 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 14:00:15.891214 134612197783360 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:00:15.891281 134612197783360 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 14:00:15.907201 134612197783360 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:00:15.907263 134612197783360 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 14:00:15.936915 134612197783360 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:00:15.937000 134612197783360 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0424 14:00:15.957010 134612197783360 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:00:15.957072 134612197783360 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0424 14:00:31.388062 134612197783360 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. I0424 14:00:31.390421 134612197783360 metric_logger.py:301] number parameters: 1.104 billion I0424 14:00:32.333340 134612197783360 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 14:00:46.817961 134612197783360 checkpointing.py:798] Waited 14.48457932472229 seconds for step 0 to finish before starting checkpointing. I0424 14:00:46.820124 134612197783360 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:00:46.822062 134612197783360 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0424 14:00:46.823543 134612197783360 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_09_pdb_lt_1/checkpoints/0. I0424 14:00:47.196895 134612197783360 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:00:47.197861 134612197783360 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:00:47.197919 134612197783360 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 14:00:47.463631 134612197783360 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.267507s I0424 14:00:47.463801 134612197783360 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.577 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.27660679817199707 s) (per-host) I0424 14:00:47.463854 134612197783360 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.276670s (batch_requests_ready=0.003012s, total_serialization_initiated=0.273587s, others=0.000071s) I0424 14:00:47.463968 134612197783360 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.280954s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.280936) I0424 14:00:47.464767 134612197783360 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.64 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_09_pdb_lt_1/checkpoints/0. I0424 14:00:47.465113 134483713963776 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:20:47.465077 I0424 14:00:47.491869 134612197783360 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:00:47.492213 134483697178368 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:00:47.492373 134612197783360 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_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777039246.8201058, 'wait_for_prev_duration_secs': 6.222724914550781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777039246.8221016, 'checkpointer_blocking_duration_secs': 0.6431658267974854, 'get_old_steps_start_time': 1777039247.4652915, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1777039246.818355, 'checkpoint_manager_blocking_duration_secs': 0.6739785671234131} I0424 14:00:47.492484 134612197783360 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 14:00:47.492553 134612197783360 max_utils.py:750] Memstats: After params initialized: I0424 14:00:47.492608 134612197783360 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0424 14:00:47.492643 134612197783360 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0424 14:00:47.492672 134612197783360 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0424 14:00:47.492698 134612197783360 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0424 14:00:47.804495 134612197783360 metric_logger.py:196] completed step: 0, seconds: 0.942, TFLOP/s/device: 3.607, Tokens/s/device: 543.628, total_weights: 16384, loss: 10.864, lm_loss: 10.864, perplexity: 52258.910 I0424 14:00:47.974522 134612197783360 metric_logger.py:196] completed step: 1, seconds: 15.469, TFLOP/s/device: 0.220, Tokens/s/device: 33.098, total_weights: 16384, loss: 10.864, lm_loss: 10.864, perplexity: 52258.910 I0424 14:00:48.381489 134612197783360 metric_logger.py:196] completed step: 2, seconds: 0.114, TFLOP/s/device: 29.714, Tokens/s/device: 4478.813, total_weights: 16384, loss: 9.760, lm_loss: 9.760, perplexity: 17319.064 I0424 14:00:48.440543 134612197783360 metric_logger.py:196] completed step: 3, seconds: 0.395, TFLOP/s/device: 8.600, Tokens/s/device: 1296.317, total_weights: 16384, loss: 8.820, lm_loss: 8.820, perplexity: 6771.344 I0424 14:00:48.558239 134612197783360 metric_logger.py:196] completed step: 4, seconds: 0.076, TFLOP/s/device: 44.889, Tokens/s/device: 6766.132, total_weights: 16384, loss: 7.947, lm_loss: 7.947, perplexity: 2825.893 I0424 14:00:48.565727 134612197783360 metric_logger.py:196] completed step: 5, seconds: 0.058, TFLOP/s/device: 58.232, Tokens/s/device: 8777.343, total_weights: 16384, loss: 7.209, lm_loss: 7.209, perplexity: 1352.040 I0424 14:00:51.109670 2537 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:00:53.143186 134486431889152 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_3 I0424 14:01:02.882992 134612197783360 metric_logger.py:196] completed step: 6, seconds: 0.119, TFLOP/s/device: 28.635, Tokens/s/device: 4316.122, total_weights: 16384, loss: 6.651, lm_loss: 6.651, perplexity: 773.347 I0424 14:01:02.941442 134612197783360 metric_logger.py:196] completed step: 7, seconds: 14.260, TFLOP/s/device: 0.238, Tokens/s/device: 35.904, total_weights: 16384, loss: 6.267, lm_loss: 6.267, perplexity: 526.951 I0424 14:01:02.999918 134612197783360 metric_logger.py:196] completed step: 8, seconds: 0.064, TFLOP/s/device: 53.280, Tokens/s/device: 8030.995, total_weights: 16384, loss: 6.030, lm_loss: 6.030, perplexity: 415.600 I0424 14:01:03.059147 134612197783360 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 14:01:03.059804 134612197783360 checkpointing.py:798] Waited 0.0006642341613769531 seconds for step 9 to finish before starting checkpointing. I0424 14:01:03.061900 134612197783360 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:01:23.688451 134483713963776 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 43.275 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.50120663642883 s) (per-host) I0424 14:01:23.688580 134483713963776 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.223355s. I0424 14:01:30.471139 134483713963776 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 43.005896s. I0424 14:01:30.471442 134483697178368 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:01:30.471583 134483697178368 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:01:30.471635 134483697178368 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:01:30.472990 134483697178368 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 14:01:30.473158 134612197783360 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0424 14:01:30.473297 134612197783360 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.411388 seconds. If this number is high, consider checkpointing less frequently. I0424 14:01:30.475064 134612197783360 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0424 14:01:30.477114 134612197783360 event_tracking.py:70] [process=3] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_09_pdb_lt_1/checkpoints/9. I0424 14:01:30.770310 134612197783360 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:01:30.770401 134612197783360 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 14:01:30.813766 134612197783360 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044929s I0424 14:01:30.813954 134612197783360 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 29.602 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.052110910415649414 s) (per-host) I0424 14:01:30.814009 134612197783360 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.052188s (batch_requests_ready=0.002664s, total_serialization_initiated=0.049438s, others=0.000086s) I0424 14:01:30.814109 134612197783360 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.056093s (all_items=0.000015s, per_item={'items': '0.00001454'}, temp_paths=0.056078) I0424 14:01:30.814818 134612197783360 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.34 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_09_pdb_lt_1/checkpoints/9. I0424 14:01:30.815173 134483697178368 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:21:30.815137 I0424 14:01:30.817069 134612197783360 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:01:30.817297 134486431889152 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:01:30.817433 134612197783360 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_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777039263.0618718, 'wait_for_prev_duration_secs': 27.41138792037964, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777039290.475103, 'checkpointer_blocking_duration_secs': 0.34021830558776855, 'get_old_steps_start_time': 1777039290.8153417, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1777039263.0600483, 'checkpoint_manager_blocking_duration_secs': 27.757351875305176} I0424 14:01:30.817593 134612197783360 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 14:01:30.817639 134612197783360 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:01:35.951703 134479410603776 array_metadata_store.py:203] [process=3][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_3 I0424 14:02:11.560955 134483697178368 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 38.716 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.79908990859985 s) (per-host) I0424 14:02:11.561057 134483697178368 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.745771s. I0424 14:02:19.444164 134483697178368 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 48.628862s. I0424 14:02:19.444441 134486431889152 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:02:19.444575 134486431889152 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:02:19.444628 134486431889152 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:02:19.446271 134486431889152 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 14:02:19.446461 134612197783360 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0424 14:02:19.446641 134612197783360 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:02:19.447554 134612197783360 metric_logger.py:196] completed step: 9, seconds: 0.059, TFLOP/s/device: 57.926, Tokens/s/device: 8731.241, 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: Fri Apr 24 14:02:30 UTC 2026 EXIT_CODE=0