XPK Start: Fri Apr 24 14:31:24 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. `rope_parameters`'s factor field must be a float >= 1, got 40 `rope_parameters`'s beta_fast field must be a float, got 32 `rope_parameters`'s beta_slow field must be a float, got 1 DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 2026-04-24 14:31:49.902656: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303) I0424 14:31:50.115584 140269606082368 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:31:59.158396 140269606082368 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:31:59.160870 140269606082368 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-sv-91ujc-slice-job-0-0.mt-15-ocdbt-false-sv-91ujc:8482 I0424 14:32:00.270560 140269606082368 max_utils.py:284] Jax distributed system initialized! I0424 14:32:06.471874 140269606082368 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:32:06.471981 140269606082368 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:32:06.472023 140269606082368 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 14:32:06.472059 140269606082368 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:32:07.193828 140269606082368 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:32:07.312020 140269606082368 checkpointing.py:688] Setting up checkpoint logger... I0424 14:32:07.312161 140269606082368 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0424 14:32:07.312208 140269606082368 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:32:07.312428 140269606082368 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 0x7f925b7757c0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:32:10.209772 140269606082368 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:32:10.210011 140269606082368 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f7e906611c0>}, handler_registry=None I0424 14:32:10.210265 140269606082368 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f7e906611c0>` 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 14:32:10.210321 140269606082368 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f7e9050b440>` 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 14:32:10.210359 140269606082368 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 0x7f7e906611c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f7e906611c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f7e9050b440>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f7e9050b440>}). I0424 14:32:10.210690 140269606082368 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:32:10.210764 140269606082368 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7f7e506cb560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:32:10.971732 140269606082368 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_15_ocdbt_false/checkpoints I0424 14:32:11.354082 140269606082368 checkpoint_manager.py:929] [process=6][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=5), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f7e9040c5f0> I0424 14:32:11.354282 140269606082368 checkpointing.py:302] Checkpoint manager created! I0424 14:32:12.327088 140269606082368 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:32:12.327223 140269606082368 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 14:32:13.727241 140269606082368 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:32:13.727363 140269606082368 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:32:13.727415 140269606082368 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:32:13.727458 140269606082368 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:32:13.727495 140269606082368 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:32:13.727529 140269606082368 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:32:13.727571 140269606082368 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:32:13.727609 140269606082368 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:32:13.727645 140269606082368 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:32:13.727678 140269606082368 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:32:13.727710 140269606082368 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:32:13.727742 140269606082368 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:32:14.134320 140269606082368 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:32:14.134417 140269606082368 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:32:14.139824 140269606082368 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:32:14.139883 140269606082368 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:32:14.156398 140269606082368 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:32:14.156466 140269606082368 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:32:14.172367 140269606082368 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:32:14.172428 140269606082368 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:32:14.197602 140269606082368 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:32:14.197671 140269606082368 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:32:14.213827 140269606082368 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:32:14.213890 140269606082368 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:32:14.229740 140269606082368 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 14:32:14.229806 140269606082368 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:32:14.259649 140269606082368 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:32:14.259721 140269606082368 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:32:14.280040 140269606082368 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:32:14.280117 140269606082368 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:32:25.455788 140269606082368 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0424 14:32:25.458235 140269606082368 metric_logger.py:301] number parameters: 1.104 billion I0424 14:32:36.704782 140269606082368 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 14:32:36.857787 140269606082368 checkpointing.py:798] Waited 0.15299081802368164 seconds for step 0 to finish before starting checkpointing. I0424 14:32:36.860115 140269606082368 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:32:36.861711 140269606082368 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0424 14:32:36.863202 140269606082368 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/0. I0424 14:32:37.157748 140269606082368 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:32:37.158694 140269606082368 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:32:37.158754 140269606082368 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 14:32:37.432900 140269606082368 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.275931s I0424 14:32:37.433070 140269606082368 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.410 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28511500358581543 s) (per-host) I0424 14:32:37.433142 140269606082368 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285197s (batch_requests_ready=0.003180s, total_serialization_initiated=0.281927s, others=0.000090s) I0424 14:32:37.433242 140269606082368 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289235s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.289218) I0424 14:32:37.434081 140269606082368 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.57 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_15_ocdbt_false/checkpoints/0. I0424 14:32:37.434395 140141955127040 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:52:37.434364 I0424 14:32:37.454003 140269606082368 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:32:37.454309 140141377554176 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:32:37.454437 140269606082368 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041156.8600852, 'wait_for_prev_duration_secs': 7.081031799316406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041156.8617496, 'checkpointer_blocking_duration_secs': 0.5728006362915039, 'get_old_steps_start_time': 1777041157.4345813, 'get_old_steps_duration_secs': 3.218650817871094e-05, 'checkpoint_manager_blocking_start_time': 1777041156.8582091, 'checkpoint_manager_blocking_duration_secs': 0.5961875915527344} I0424 14:32:37.454544 140269606082368 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 14:32:37.454614 140269606082368 max_utils.py:750] Memstats: After params initialized: I0424 14:32:37.454665 140269606082368 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 14:32:37.454697 140269606082368 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 14:32:37.454724 140269606082368 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 14:32:37.454748 140269606082368 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 14:32:37.770616 140269606082368 metric_logger.py:196] completed step: 0, seconds: 11.245, TFLOP/s/device: 1.208, Tokens/s/device: 182.118, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:32:37.941446 140269606082368 metric_logger.py:196] completed step: 1, seconds: 1.064, TFLOP/s/device: 12.770, Tokens/s/device: 1924.839, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:32:38.400734 140269606082368 metric_logger.py:196] completed step: 2, seconds: 0.025, TFLOP/s/device: 549.841, Tokens/s/device: 82878.070, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0424 14:32:38.548360 140269606082368 metric_logger.py:196] completed step: 3, seconds: 0.449, TFLOP/s/device: 30.256, Tokens/s/device: 4560.485, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0424 14:32:38.554110 140269606082368 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0424 14:32:38.843243 140269606082368 checkpointing.py:798] Waited 0.28911638259887695 seconds for step 5 to finish before starting checkpointing. I0424 14:32:38.845976 140269606082368 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:32:40.554758 2783 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:32:42.577277 140141385946880 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/0/items/array_metadatas/process_6 I0424 14:32:47.683198 140141955127040 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 149.935 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.535208463668823 s) (per-host) I0424 14:32:47.683316 140141955127040 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.248809s. I0424 14:32:54.663999 140141955127040 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.229475s. I0424 14:32:54.664252 140141377554176 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:32:54.664325 140141377554176 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:32:54.664389 140141377554176 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:32:54.666988 140141377554176 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 14:32:54.667186 140269606082368 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0424 14:32:54.667322 140269606082368 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.821345 seconds. If this number is high, consider checkpointing less frequently. I0424 14:32:54.669326 140269606082368 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0424 14:32:54.671960 140269606082368 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/5. I0424 14:32:54.966748 140269606082368 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:32:54.966842 140269606082368 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 14:32:55.004151 140269606082368 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.038917s I0424 14:32:55.004316 140269606082368 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 33.469 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.046090126037597656 s) (per-host) I0424 14:32:55.004368 140269606082368 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.046153s (batch_requests_ready=0.002759s, total_serialization_initiated=0.043324s, others=0.000070s) I0424 14:32:55.004466 140269606082368 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051163s (all_items=0.000015s, per_item={'items': '0.00001478'}, temp_paths=0.051148) I0424 14:32:55.005228 140269606082368 event_tracking.py:125] [process=6] [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_15_ocdbt_false/checkpoints/5. I0424 14:32:55.005527 140141377554176 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:52:55.005494 I0424 14:32:55.020415 140269606082368 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:32:55.020701 140140890552064 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:32:55.020859 140269606082368 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041158.8459468, 'wait_for_prev_duration_secs': 15.82134461402893, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777041174.669369, 'checkpointer_blocking_duration_secs': 0.33629727363586426, 'get_old_steps_start_time': 1777041175.0056877, 'get_old_steps_duration_secs': 2.9325485229492188e-05, 'checkpoint_manager_blocking_start_time': 1777041158.8436012, 'checkpoint_manager_blocking_duration_secs': 16.1772243976593} I0424 14:32:55.020972 140269606082368 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0424 14:32:55.022023 140269606082368 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.696, Tokens/s/device: 12464.851, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0424 14:32:55.029769 140269606082368 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.303, Tokens/s/device: 13912.854, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0424 14:33:00.620496 140141385946880 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/5/items/array_metadatas/process_6 I0424 14:33:05.561370 140141377554176 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 148.975 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.603115558624268 s) (per-host) I0424 14:33:05.561480 140141377554176 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.555841s. I0424 14:33:12.181410 140141377554176 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.175756s. I0424 14:33:12.181705 140140890552064 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:33:12.181834 140140890552064 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:33:12.181885 140140890552064 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:33:12.183228 140140890552064 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0424 14:33:18.105825 140269606082368 metric_logger.py:196] completed step: 6, seconds: 16.474, TFLOP/s/device: 0.825, Tokens/s/device: 124.320, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0424 14:33:18.253512 140269606082368 metric_logger.py:196] completed step: 7, seconds: 22.931, TFLOP/s/device: 0.593, Tokens/s/device: 89.312, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0424 14:33:18.401115 140269606082368 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.629, Tokens/s/device: 13359.165, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0424 14:33:18.549258 140269606082368 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 14:33:18.549952 140269606082368 checkpointing.py:798] Waited 0.0007050037384033203 seconds for step 9 to finish before starting checkpointing. I0424 14:33:18.551812 140269606082368 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:33:18.553373 140269606082368 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0424 14:33:18.554844 140269606082368 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9. I0424 14:33:18.864192 140269606082368 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:33:18.864350 140269606082368 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 14:33:18.900662 140269606082368 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.038074s I0424 14:33:18.900813 140269606082368 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.114 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04521799087524414 s) (per-host) I0424 14:33:18.900863 140269606082368 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045277s (batch_requests_ready=0.002885s, total_serialization_initiated=0.042329s, others=0.000064s) I0424 14:33:18.900954 140269606082368 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049405s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.049394) I0424 14:33:18.901660 140269606082368 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9. I0424 14:33:18.902004 140140320110336 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:53:18.901968 I0424 14:33:18.903826 140269606082368 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:33:18.904032 140134543582976 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:33:18.904138 140269606082368 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_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777041198.551784, 'wait_for_prev_duration_secs': 6.914138793945312e-05, 'time_between_consecutive_saves_sec': 6.368512153625488, 'checkpointer_blocking_start_time': 1777041198.5534096, 'checkpointer_blocking_duration_secs': 0.3487050533294678, 'get_old_steps_start_time': 1777041198.902133, 'get_old_steps_duration_secs': 2.384185791015625e-05, 'checkpoint_manager_blocking_start_time': 1777041198.5501955, 'checkpoint_manager_blocking_duration_secs': 0.3539137840270996} I0424 14:33:18.904295 140269606082368 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 14:33:18.904341 140269606082368 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:33:24.185153 140140843513600 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/nnx_xpk_feat_nnx_post_train_fixes_20260424_120657_15_ocdbt_false/checkpoints/9/items/array_metadatas/process_6 I0424 14:33:29.517660 140140320110336 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 148.152 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.66201901435852 s) (per-host) I0424 14:33:29.517775 140140320110336 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.615166s. I0424 14:33:36.116962 140140320110336 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 17.214337s. I0424 14:33:36.117204 140134543582976 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:33:36.117269 140134543582976 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:33:36.117310 140134543582976 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:33:36.119541 140134543582976 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 14:33:36.119711 140269606082368 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0424 14:33:36.119883 140269606082368 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:33:36.120858 140269606082368 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.841, Tokens/s/device: 13843.263, total_weights: 65536, loss: 7.176, lm_loss: 7.176, perplexity: 1307.366 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Fri Apr 24 14:33:44 UTC 2026 EXIT_CODE=0