XPK Start: Fri Apr 24 14:21:19 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 14:21:45.086755: 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 14:21:45.299365 137912562218816 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 14:21:54.340158 137912562218816 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 14:21:54.344718 137912562218816 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-tfnus-slice-job-0-0.mt-15-ocdbt-true-sv-tfnus:8482 I0424 14:21:54.785536 137912562218816 max_utils.py:284] Jax distributed system initialized! I0424 14:21:59.939519 137912562218816 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 14:21:59.939625 137912562218816 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 14:21:59.939671 137912562218816 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 14:21:59.939721 137912562218816 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 14:22:00.633594 137912562218816 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 14:22:00.748136 137912562218816 checkpointing.py:688] Setting up checkpoint logger... I0424 14:22:00.748257 137912562218816 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0424 14:22:00.748301 137912562218816 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 14:22:00.748503 137912562218816 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 0x7d6dae518410>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 14:22:03.709183 137912562218816 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 14:22:03.709422 137912562218816 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 0x7d5a2040cef0>}, handler_registry=None I0424 14:22:03.709691 137912562218816 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d5a2040cef0>` 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 14:22:03.709747 137912562218816 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d5a2044c7d0>` 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 14:22:03.709782 137912562218816 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 0x7d5a2040cef0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d5a2040cef0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d5a2044c7d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d5a2044c7d0>}). I0424 14:22:03.710120 137912562218816 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 14:22:03.710190 137912562218816 async_checkpointer.py:192] [process=3][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d5a20447560> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 14:22:04.878793 137912562218816 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_true/checkpoints I0424 14:22:04.899497 137912562218816 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=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_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d59f07c4290> I0424 14:22:04.899664 137912562218816 checkpointing.py:302] Checkpoint manager created! I0424 14:22:05.875000 137912562218816 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 14:22:05.875125 137912562218816 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0424 14:22:07.266361 137912562218816 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0424 14:22:07.266502 137912562218816 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:22:07.266579 137912562218816 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0424 14:22:07.266647 137912562218816 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0424 14:22:07.266707 137912562218816 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:22:07.266766 137912562218816 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0424 14:22:07.266821 137912562218816 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:22:07.266875 137912562218816 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0424 14:22:07.266942 137912562218816 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:22:07.266999 137912562218816 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0424 14:22:07.267054 137912562218816 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0424 14:22:07.267106 137912562218816 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0424 14:22:07.673801 137912562218816 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:22:07.673897 137912562218816 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:22:07.679377 137912562218816 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 14:22:07.679442 137912562218816 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:22:07.696186 137912562218816 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:22:07.696249 137912562218816 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:22:07.712210 137912562218816 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 14:22:07.712270 137912562218816 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 14:22:07.737906 137912562218816 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 14:22:07.737994 137912562218816 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:22:07.753983 137912562218816 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 14:22:07.754043 137912562218816 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:22:07.769886 137912562218816 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 14:22:07.769961 137912562218816 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:22:07.799918 137912562218816 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 14:22:07.800005 137912562218816 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 14:22:07.820033 137912562218816 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 14:22:07.820098 137912562218816 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 14:22:19.324187 137912562218816 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 14:22:19.326514 137912562218816 metric_logger.py:301] number parameters: 1.104 billion I0424 14:22:31.013092 137912562218816 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0424 14:22:31.166372 137912562218816 checkpointing.py:798] Waited 0.15326309204101562 seconds for step 0 to finish before starting checkpointing. I0424 14:22:31.168703 137912562218816 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:22:31.170513 137912562218816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 0 I0424 14:22:31.172233 137912562218816 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_15_ocdbt_true/checkpoints/0. I0424 14:22:31.512988 137912562218816 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 14:22:31.513988 137912562218816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:22:31.514046 137912562218816 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 14:22:31.792905 137912562218816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.280699s I0424 14:22:31.793099 137912562218816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 5.324 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2897467613220215 s) (per-host) I0424 14:22:31.793151 137912562218816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.289810s (batch_requests_ready=0.002969s, total_serialization_initiated=0.286768s, others=0.000073s) I0424 14:22:31.793251 137912562218816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.293993s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.293974) I0424 14:22:31.794104 137912562218816 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.62 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_true/checkpoints/0. I0424 14:22:31.794447 137785820854016 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:42:31.794408 I0424 14:22:31.797651 137912562218816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:22:31.797910 137784217126656 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:22:31.798077 137912562218816 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040551.1686852, 'wait_for_prev_duration_secs': 6.67572021484375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777040551.170602, 'checkpointer_blocking_duration_secs': 0.6239550113677979, 'get_old_steps_start_time': 1777040551.7945848, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1777040551.166777, 'checkpoint_manager_blocking_duration_secs': 0.6312613487243652} I0424 14:22:31.798191 137912562218816 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0424 14:22:31.798262 137912562218816 max_utils.py:750] Memstats: After params initialized: I0424 14:22:31.798310 137912562218816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_10(process=3,(2,2,0,0)) I0424 14:22:31.798341 137912562218816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_11(process=3,(3,2,0,0)) I0424 14:22:31.798367 137912562218816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_14(process=3,(2,3,0,0)) I0424 14:22:31.798390 137912562218816 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_15(process=3,(3,3,0,0)) I0424 14:22:32.115519 137912562218816 metric_logger.py:196] completed step: 0, seconds: 11.685, TFLOP/s/device: 1.163, Tokens/s/device: 175.260, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:22:32.299970 137912562218816 metric_logger.py:196] completed step: 1, seconds: 1.101, TFLOP/s/device: 12.345, Tokens/s/device: 1860.758, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0424 14:22:32.734357 137912562218816 metric_logger.py:196] completed step: 2, seconds: 0.038, TFLOP/s/device: 356.627, Tokens/s/device: 53754.692, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0424 14:22:32.881705 137912562218816 metric_logger.py:196] completed step: 3, seconds: 0.424, TFLOP/s/device: 32.025, Tokens/s/device: 4827.206, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0424 14:22:32.887547 137912562218816 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0424 14:22:33.177356 137912562218816 checkpointing.py:798] Waited 0.2897791862487793 seconds for step 5 to finish before starting checkpointing. I0424 14:22:33.180182 137912562218816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:22:34.802823 2749 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 14:22:36.870382 137784225519360 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_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_3 I0424 14:23:08.359587 137785820854016 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 42.858 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.85619807243347 s) (per-host) I0424 14:23:08.359721 137785820854016 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.564617s. I0424 14:23:16.639602 137785820854016 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 44.844481s. I0424 14:23:16.639887 137784217126656 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:23:16.640041 137784217126656 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:23:16.640093 137784217126656 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:23:16.642112 137784217126656 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0424 14:23:16.642289 137912562218816 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 14:23:16.642420 137912562218816 checkpoint_manager.py:1452] Waiting for previous save to complete took 43.462239 seconds. If this number is high, consider checkpointing less frequently. I0424 14:23:16.644436 137912562218816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 5 I0424 14:23:16.647049 137912562218816 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_15_ocdbt_true/checkpoints/5. I0424 14:23:16.937608 137912562218816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:23:16.937699 137912562218816 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 14:23:16.974188 137912562218816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.038212s I0424 14:23:16.974353 137912562218816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 34.122 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04520726203918457 s) (per-host) I0424 14:23:16.974405 137912562218816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045269s (batch_requests_ready=0.002589s, total_serialization_initiated=0.042613s, others=0.000067s) I0424 14:23:16.974499 137912562218816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.050059s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.050042) I0424 14:23:16.975212 137912562218816 event_tracking.py:125] [process=3] [async] Finished blocking save in 0.33 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_true/checkpoints/5. I0424 14:23:16.975501 137784217126656 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:43:16.975467 I0424 14:23:16.993438 137912562218816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:23:16.993682 137764498536192 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:23:16.993810 137912562218816 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040553.180153, 'wait_for_prev_duration_secs': 43.462238788604736, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777040596.6444752, 'checkpointer_blocking_duration_secs': 0.3311622142791748, 'get_old_steps_start_time': 1777040596.9756618, 'get_old_steps_duration_secs': 3.4809112548828125e-05, 'checkpoint_manager_blocking_start_time': 1777040553.1777003, 'checkpoint_manager_blocking_duration_secs': 43.81607484817505} I0424 14:23:16.993918 137912562218816 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0424 14:23:16.995000 137912562218816 metric_logger.py:196] completed step: 4, seconds: 0.164, TFLOP/s/device: 82.634, Tokens/s/device: 12455.451, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0424 14:23:17.002602 137912562218816 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.422, Tokens/s/device: 13930.836, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0424 14:23:22.637260 137784225519360 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_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_3 I0424 14:23:40.073866 137912562218816 metric_logger.py:196] completed step: 6, seconds: 44.113, TFLOP/s/device: 0.308, Tokens/s/device: 46.426, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0424 14:23:40.221188 137912562218816 metric_logger.py:196] completed step: 7, seconds: 22.926, TFLOP/s/device: 0.593, Tokens/s/device: 89.331, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0424 14:23:40.368651 137912562218816 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.653, Tokens/s/device: 13362.738, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0424 14:23:40.516788 137912562218816 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0424 14:23:40.517433 137912562218816 checkpointing.py:798] Waited 0.0006587505340576172 seconds for step 9 to finish before starting checkpointing. I0424 14:23:40.519613 137912562218816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:23:59.132752 137784217126656 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.428 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.20356893539429 s) (per-host) I0424 14:23:59.132884 137784217126656 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.157274s. I0424 14:24:07.732733 137784217126656 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 50.757107s. I0424 14:24:07.732989 137764498536192 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:24:07.733052 137764498536192 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:24:07.733102 137764498536192 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:24:07.734475 137764498536192 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0424 14:24:07.734657 137912562218816 checkpoint_manager.py:2032] [process=3][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0424 14:24:07.734785 137912562218816 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.215174 seconds. If this number is high, consider checkpointing less frequently. I0424 14:24:07.736259 137912562218816 checkpoint_manager.py:1512] [process=3] Saving checkpoint at step 9 I0424 14:24:07.738409 137912562218816 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_15_ocdbt_true/checkpoints/9. I0424 14:24:08.037335 137912562218816 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0424 14:24:08.037427 137912562218816 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 14:24:08.072407 137912562218816 base_pytree_checkpoint_handler.py:154] [process=3][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.036593s I0424 14:24:08.072574 137912562218816 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/blocking_gbytes_per_sec: 35.493 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04346132278442383 s) (per-host) I0424 14:24:08.072623 137912562218816 base_pytree_checkpoint_handler.py:768] [process=3][thread=MainThread] Initiated Pytree async_save. Time taken: 0.043521s (batch_requests_ready=0.002625s, total_serialization_initiated=0.040831s, others=0.000066s) I0424 14:24:08.072727 137912562218816 composite_checkpoint_handler.py:715] [process=3][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.047544s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.047534) I0424 14:24:08.073346 137912562218816 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_15_ocdbt_true/checkpoints/9. I0424 14:24:08.073651 137783641347840 async_checkpointer.py:76] [process=3][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 14:44:08.073615 I0424 14:24:08.075396 137912562218816 checkpoint_manager.py:1560] [process=3][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0424 14:24:08.075626 137764498536192 async_checkpointer.py:280] [process=3][thread=save_finalize] Waiting for background save thread=async_save. I0424 14:24:08.075780 137912562218816 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_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777040620.5195866, 'wait_for_prev_duration_secs': 27.215174198150635, 'time_between_consecutive_saves_sec': 23.87743639945984, 'checkpointer_blocking_start_time': 1777040647.7362988, 'checkpointer_blocking_duration_secs': 0.3374972343444824, 'get_old_steps_start_time': 1777040648.0738115, 'get_old_steps_duration_secs': 2.3365020751953125e-05, 'checkpoint_manager_blocking_start_time': 1777040620.517676, 'checkpoint_manager_blocking_duration_secs': 27.558073043823242} I0424 14:24:08.075890 137912562218816 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0424 14:24:08.075947 137912562218816 checkpoint_manager.py:2020] [process=3][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 14:24:13.139716 137784225519360 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_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_3 I0424 14:24:49.772068 137783641347840 base_pytree_checkpoint_handler.py:130] [process=3] /jax/orbax/write/gbytes_per_sec: 37.841 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.74293041229248 s) (per-host) I0424 14:24:49.772174 137783641347840 async_checkpointer.py:90] [process=3][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.698408s. I0424 14:24:57.659698 137783641347840 async_checkpointer.py:160] [process=3][thread=async_save] Background save thread done. Time taken: 49.585914s. I0424 14:24:57.660019 137764498536192 async_checkpointer.py:288] [process=3][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 14:24:57.660133 137764498536192 async_checkpointer.py:298] [process=3][thread=save_finalize] No errors found in background save thread=async_save. I0424 14:24:57.660174 137764498536192 checkpoint_manager.py:2137] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0424 14:24:57.661417 137764498536192 checkpoint_manager.py:2146] [process=3][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0424 14:24:57.661602 137912562218816 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 14:24:57.661778 137912562218816 checkpoint_manager.py:2009] [process=3][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 14:24:57.662824 137912562218816 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 92.042, Tokens/s/device: 13873.553, 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 14:25:09 UTC 2026 EXIT_CODE=0