XPK Start: Sat Apr 25 14:27:50 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. 2026-04-25 14:28:15.103761: 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) I0425 14:28:15.681018 136188498442048 max_utils.py:273] Attempting to initialize the jax distributed system... I0425 14:28:24.722086 136188498442048 distributed.py:149] Starting JAX distributed service on [::]:8482 I0425 14:28:24.724573 136188498442048 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-puriz-slice-job-0-0.mt-15-ocdbt-true-sv-puriz:8482 I0425 14:28:25.692840 136188498442048 max_utils.py:284] Jax distributed system initialized! I0425 14:28:30.614708 136188498442048 max_utils.py:800] System Information: Jax Version: 0.9.2 I0425 14:28:30.614812 136188498442048 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0425 14:28:30.614854 136188498442048 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 I0425 14:28:30.614890 136188498442048 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 14:28:31.315911 136188498442048 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 14:28:31.433156 136188498442048 checkpointing.py:688] Setting up checkpoint logger... I0425 14:28:31.433285 136188498442048 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0425 14:28:31.433332 136188498442048 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0425 14:28:31.433540 136188498442048 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 0x7bdc3f91d100>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 14:28:34.396089 136188498442048 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0425 14:28:34.396357 136188498442048 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 0x7bc78c3f58b0>}, handler_registry=None I0425 14:28:34.396603 136188498442048 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bc78c3f58b0>` 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`. I0425 14:28:34.396651 136188498442048 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bc78c3f52e0>` 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`. I0425 14:28:34.396688 136188498442048 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 0x7bc78c3f58b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bc78c3f58b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bc78c3f52e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bc78c3f52e0>}). I0425 14:28:34.397009 136188498442048 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0425 14:28:34.397077 136188498442048 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7bc78c7271a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0425 14:28:35.957927 136188498442048 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints I0425 14:28:36.239494 136188498442048 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bc87c52e0c0> I0425 14:28:36.239671 136188498442048 checkpointing.py:302] Checkpoint manager created! I0425 14:28:37.207330 136188498442048 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0425 14:28:37.207453 136188498442048 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0425 14:28:38.583728 136188498442048 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0425 14:28:38.583847 136188498442048 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0425 14:28:38.583899 136188498442048 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0425 14:28:38.583942 136188498442048 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0425 14:28:38.583978 136188498442048 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0425 14:28:38.584012 136188498442048 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0425 14:28:38.584047 136188498442048 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 14:28:38.584082 136188498442048 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0425 14:28:38.584131 136188498442048 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 14:28:38.584167 136188498442048 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0425 14:28:38.584199 136188498442048 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0425 14:28:38.584231 136188498442048 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0425 14:28:38.994015 136188498442048 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 14:28:38.994119 136188498442048 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 14:28:38.999425 136188498442048 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 14:28:38.999481 136188498442048 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 14:28:39.016137 136188498442048 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 14:28:39.016196 136188498442048 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 14:28:39.037977 136188498442048 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 14:28:39.038088 136188498442048 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 14:28:39.064542 136188498442048 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 14:28:39.064611 136188498442048 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 14:28:39.080750 136188498442048 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 14:28:39.080812 136188498442048 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 14:28:39.096891 136188498442048 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 14:28:39.096953 136188498442048 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 14:28:39.126849 136188498442048 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0425 14:28:39.126920 136188498442048 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 14:28:39.147061 136188498442048 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 14:28:39.147135 136188498442048 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 14:28:50.666026 136188498442048 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. I0425 14:28:50.668509 136188498442048 metric_logger.py:301] number parameters: 1.104 billion I0425 14:29:01.935910 136188498442048 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0425 14:29:02.088849 136188498442048 checkpointing.py:798] Waited 0.1529216766357422 seconds for step 0 to finish before starting checkpointing. I0425 14:29:02.091017 136188498442048 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 14:29:02.092648 136188498442048 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0425 14:29:02.093999 136188498442048 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/0. I0425 14:29:02.424804 136188498442048 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 14:29:02.425718 136188498442048 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0425 14:29:02.425778 136188498442048 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 I0425 14:29:02.704695 136188498442048 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280713s I0425 14:29:02.704869 136188498442048 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.318 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2900657653808594 s) (per-host) I0425 14:29:02.704918 136188498442048 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290126s (batch_requests_ready=0.003080s, total_serialization_initiated=0.286978s, others=0.000067s) I0425 14:29:02.705014 136188498442048 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.294411s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.294392) I0425 14:29:02.705698 136188498442048 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.61 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/0. I0425 14:29:02.705925 136061167015680 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 14:49:02.705889 I0425 14:29:02.711832 136188498442048 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0425 14:29:02.712065 136062236399360 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 14:29:02.712198 136188498442048 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777127342.0910003, 'wait_for_prev_duration_secs': 5.841255187988281e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777127342.0926874, 'checkpointer_blocking_duration_secs': 0.6133828163146973, 'get_old_steps_start_time': 1777127342.7061045, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1777127342.089257, 'checkpoint_manager_blocking_duration_secs': 0.6229045391082764} I0425 14:29:02.712299 136188498442048 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0425 14:29:02.712357 136188498442048 max_utils.py:750] Memstats: After params initialized: I0425 14:29:02.712405 136188498442048 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0425 14:29:02.712436 136188498442048 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0425 14:29:02.712463 136188498442048 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0425 14:29:02.712486 136188498442048 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0425 14:29:03.031396 136188498442048 metric_logger.py:196] completed step: 0, seconds: 11.266, TFLOP/s/device: 1.206, Tokens/s/device: 181.782, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0425 14:29:03.218643 136188498442048 metric_logger.py:196] completed step: 1, seconds: 1.094, TFLOP/s/device: 12.424, Tokens/s/device: 1872.673, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0425 14:29:04.010007 136188498442048 metric_logger.py:196] completed step: 2, seconds: 0.041, TFLOP/s/device: 329.313, Tokens/s/device: 49637.655, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0425 14:29:04.157185 136188498442048 metric_logger.py:196] completed step: 3, seconds: 0.781, TFLOP/s/device: 17.396, Tokens/s/device: 2622.081, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0425 14:29:04.163252 136188498442048 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0425 14:29:04.452168 136188498442048 checkpointing.py:798] Waited 0.2888524532318115 seconds for step 5 to finish before starting checkpointing. I0425 14:29:04.455349 136188498442048 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 14:29:05.646513 2766 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 14:29:07.905532 136061687404288 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_6 I0425 14:29:38.770004 136061167015680 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 43.449 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.355162143707275 s) (per-host) I0425 14:29:38.770139 136061167015680 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.064105s. I0425 14:29:45.212687 136061167015680 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 42.506635s. I0425 14:29:45.212882 136062236399360 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 14:29:45.212952 136062236399360 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 14:29:45.213002 136062236399360 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0425 14:29:45.215497 136062236399360 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0425 14:29:45.215684 136188498442048 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. W0425 14:29:45.215817 136188498442048 checkpoint_manager.py:1452] Waiting for previous save to complete took 40.760468 seconds. If this number is high, consider checkpointing less frequently. I0425 14:29:45.217707 136188498442048 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0425 14:29:45.220373 136188498442048 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/5. I0425 14:29:45.524842 136188498442048 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0425 14:29:45.524935 136188498442048 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 I0425 14:29:45.564785 136188498442048 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041576s I0425 14:29:45.564946 136188498442048 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.441 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04906201362609863 s) (per-host) I0425 14:29:45.564994 136188498442048 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049119s (batch_requests_ready=0.002663s, total_serialization_initiated=0.046393s, others=0.000063s) I0425 14:29:45.565080 136188498442048 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.054019s (all_items=0.000015s, per_item={'items': '0.00001478'}, temp_paths=0.054004) I0425 14:29:45.565806 136188498442048 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/5. I0425 14:29:45.566160 136062236399360 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 14:49:45.566122 I0425 14:29:45.579514 136188498442048 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0425 14:29:45.579762 136034767599360 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 14:29:45.579921 136188498442048 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777127344.45532, 'wait_for_prev_duration_secs': 40.76046824455261, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777127385.2177458, 'checkpointer_blocking_duration_secs': 0.34856343269348145, 'get_old_steps_start_time': 1777127385.5663378, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1777127344.452527, 'checkpoint_manager_blocking_duration_secs': 41.12736105918884} I0425 14:29:45.580033 136188498442048 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0425 14:29:45.581123 136188498442048 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.623, Tokens/s/device: 12453.785, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0425 14:29:45.589172 136188498442048 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.485, Tokens/s/device: 13940.318, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0425 14:29:50.705040 136061687404288 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_6 I0425 14:30:08.915153 136188498442048 metric_logger.py:196] completed step: 6, seconds: 41.424, TFLOP/s/device: 0.328, Tokens/s/device: 49.440, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0425 14:30:09.062797 136188498442048 metric_logger.py:196] completed step: 7, seconds: 23.181, TFLOP/s/device: 0.586, Tokens/s/device: 88.349, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0425 14:30:09.210464 136188498442048 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.649, Tokens/s/device: 13362.128, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0425 14:30:09.358563 136188498442048 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0425 14:30:09.359232 136188498442048 checkpointing.py:798] Waited 0.0006816387176513672 seconds for step 9 to finish before starting checkpointing. I0425 14:30:09.361201 136188498442048 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 14:30:27.450244 136062236399360 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.668 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.934324502944946 s) (per-host) I0425 14:30:27.450355 136062236399360 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.884080s. I0425 14:30:34.077092 136062236399360 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.510802s. I0425 14:30:34.077413 136034767599360 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 14:30:34.077533 136034767599360 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 14:30:34.077589 136034767599360 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0425 14:30:34.078907 136034767599360 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0425 14:30:34.079082 136188498442048 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0425 14:30:34.079217 136188498442048 checkpoint_manager.py:1452] Waiting for previous save to complete took 24.718024 seconds. If this number is high, consider checkpointing less frequently. I0425 14:30:34.080860 136188498442048 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0425 14:30:34.082958 136188498442048 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/9. I0425 14:30:34.487509 136188498442048 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0425 14:30:34.487601 136188498442048 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 I0425 14:30:34.526756 136188498442048 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041038s I0425 14:30:34.526917 136188498442048 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 32.048 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04813337326049805 s) (per-host) I0425 14:30:34.526965 136188498442048 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048191s (batch_requests_ready=0.002788s, total_serialization_initiated=0.045341s, others=0.000062s) I0425 14:30:34.527060 136188498442048 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.052145s (all_items=0.000010s, per_item={'items': '0.00001049'}, temp_paths=0.052135) I0425 14:30:34.527801 136188498442048 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.45 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/9. I0425 14:30:34.528173 136034767599360 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 14:50:34.528137 I0425 14:30:34.530114 136188498442048 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0425 14:30:34.530422 136061687404288 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 14:30:34.530551 136188498442048 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777127409.3611712, 'wait_for_prev_duration_secs': 24.718023777008057, 'time_between_consecutive_saves_sec': 24.145633935928345, 'checkpointer_blocking_start_time': 1777127434.0808992, 'checkpointer_blocking_duration_secs': 0.44737815856933594, 'get_old_steps_start_time': 1777127434.5282938, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1777127409.3594673, 'checkpoint_manager_blocking_duration_secs': 25.171053409576416} I0425 14:30:34.530714 136188498442048 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0425 14:30:34.530757 136188498442048 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 14:30:39.562994 136059572700928 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_20260425_121355/nnx_xpk_feat_nnx_post_train_fixes_20260425_121355_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_6 I0425 14:31:16.308676 136034767599360 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.762 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.82985186576843 s) (per-host) I0425 14:31:16.308802 136034767599360 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.780020s. I0425 14:31:23.027157 136034767599360 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.498358s. I0425 14:31:23.027462 136061687404288 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 14:31:23.027593 136061687404288 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 14:31:23.027634 136061687404288 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0425 14:31:23.030133 136061687404288 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0425 14:31:23.031520 136188498442048 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. I0425 14:31:23.031683 136188498442048 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 14:31:23.032690 136188498442048 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.871, Tokens/s/device: 13847.849, 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: Sat Apr 25 14:31:35 UTC 2026 EXIT_CODE=0