XPK Start: Thu Apr 23 15:43:06 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. 2026-04-23 15:43:31.719160: 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) I0423 15:43:31.932758 136127590459200 max_utils.py:273] Attempting to initialize the jax distributed system... I0423 15:43:40.974573 136127590459200 distributed.py:149] Starting JAX distributed service on [::]:8482 I0423 15:43:40.976965 136127590459200 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-sv-eoaff-slice-job-0-0.mt-16-zarr3-false-sv-eoaff:8482 I0423 15:43:42.038646 136127590459200 max_utils.py:284] Jax distributed system initialized! I0423 15:43:47.179567 136127590459200 max_utils.py:800] System Information: Jax Version: 0.9.2 I0423 15:43:47.179674 136127590459200 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0423 15:43:47.179716 136127590459200 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 I0423 15:43:47.179753 136127590459200 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 15:43:47.873177 136127590459200 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 15:43:47.987797 136127590459200 checkpointing.py:688] Setting up checkpoint logger... I0423 15:43:47.987917 136127590459200 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0423 15:43:47.987963 136127590459200 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0423 15:43:47.988170 136127590459200 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7bce16314260>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 15:43:50.896902 136127590459200 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0423 15:43:50.897158 136127590459200 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 0x7bce6e5cf3b0>}, handler_registry=None I0423 15:43:50.897423 136127590459200 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bce6e5cf3b0>` 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`. I0423 15:43:50.897479 136127590459200 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb974631640>` 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`. I0423 15:43:50.897519 136127590459200 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 0x7bce6e5cf3b0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7bce6e5cf3b0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb974631640>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7bb974631640>}). I0423 15:43:50.897843 136127590459200 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0423 15:43:50.897912 136127590459200 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7bb9746236a0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0423 15:43:52.124447 136127590459200 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints I0423 15:43:52.131588 136127590459200 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7bb974618800> I0423 15:43:52.131701 136127590459200 checkpointing.py:302] Checkpoint manager created! I0423 15:43:53.094758 136127590459200 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0423 15:43:53.094879 136127590459200 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0423 15:43:54.558718 136127590459200 maxtext_utils.py:1889] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0423 15:43:54.558839 136127590459200 maxtext_utils.py:1889] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 15:43:54.558892 136127590459200 maxtext_utils.py:1889] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0423 15:43:54.558935 136127590459200 maxtext_utils.py:1889] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0423 15:43:54.558972 136127590459200 maxtext_utils.py:1889] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 15:43:54.559006 136127590459200 maxtext_utils.py:1889] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0423 15:43:54.559043 136127590459200 maxtext_utils.py:1889] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:43:54.559079 136127590459200 maxtext_utils.py:1889] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0423 15:43:54.559130 136127590459200 maxtext_utils.py:1889] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:43:54.559166 136127590459200 maxtext_utils.py:1889] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0423 15:43:54.559200 136127590459200 maxtext_utils.py:1889] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0423 15:43:54.559233 136127590459200 maxtext_utils.py:1889] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0423 15:43:54.966197 136127590459200 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 15:43:54.966302 136127590459200 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:43:54.971535 136127590459200 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 15:43:54.971592 136127590459200 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:43:54.988337 136127590459200 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 15:43:54.988398 136127590459200 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:43:55.004451 136127590459200 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 15:43:55.004511 136127590459200 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0423 15:43:55.029856 136127590459200 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:43:55.029927 136127590459200 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:43:55.046213 136127590459200 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:43:55.046281 136127590459200 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:43:55.062463 136127590459200 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 15:43:55.062522 136127590459200 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:43:55.092226 136127590459200 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 15:43:55.092304 136127590459200 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0423 15:43:55.112549 136127590459200 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 15:43:55.112613 136127590459200 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0423 15:44:06.556627 136127590459200 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. I0423 15:44:06.558898 136127590459200 metric_logger.py:301] number parameters: 1.104 billion I0423 15:44:18.121609 136127590459200 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0423 15:44:18.274475 136127590459200 checkpointing.py:798] Waited 0.15285062789916992 seconds for step 0 to finish before starting checkpointing. I0423 15:44:18.276877 136127590459200 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 15:44:18.278567 136127590459200 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0423 15:44:18.279972 136127590459200 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/0. I0423 15:44:18.618596 136127590459200 signaling_client.py:364] Using JaxDistributedSignalingClient I0423 15:44:18.619558 136127590459200 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:44:18.619640 136127590459200 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 I0423 15:44:18.892068 136127590459200 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.274359s I0423 15:44:18.892252 136127590459200 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.432 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28398704528808594 s) (per-host) I0423 15:44:18.892367 136127590459200 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284113s (batch_requests_ready=0.003329s, total_serialization_initiated=0.280650s, others=0.000134s) I0423 15:44:18.892468 136127590459200 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288293s (all_items=0.000019s, per_item={'items': '0.00001884'}, temp_paths=0.288274) I0423 15:44:18.893262 136127590459200 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/0. I0423 15:44:18.893592 136001737815808 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:04:18.893556 I0423 15:44:18.900604 136127590459200 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:44:18.900911 136000101250816 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:44:18.901046 136127590459200 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959058.2768588, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776959058.2786067, 'checkpointer_blocking_duration_secs': 0.6151416301727295, 'get_old_steps_start_time': 1776959058.893775, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776959058.2748826, 'checkpoint_manager_blocking_duration_secs': 0.6261260509490967} I0423 15:44:18.901213 136127590459200 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0423 15:44:18.901284 136127590459200 max_utils.py:750] Memstats: After params initialized: I0423 15:44:18.901340 136127590459200 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0423 15:44:18.901372 136127590459200 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0423 15:44:18.901398 136127590459200 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0423 15:44:18.901423 136127590459200 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0423 15:44:19.218032 136127590459200 metric_logger.py:196] completed step: 0, seconds: 11.562, TFLOP/s/device: 1.175, Tokens/s/device: 177.138, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0423 15:44:19.390176 136127590459200 metric_logger.py:196] completed step: 1, seconds: 1.094, TFLOP/s/device: 12.414, Tokens/s/device: 1871.246, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0423 15:44:19.809159 136127590459200 metric_logger.py:196] completed step: 2, seconds: 0.026, TFLOP/s/device: 520.022, Tokens/s/device: 78383.344, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0423 15:44:19.956659 136127590459200 metric_logger.py:196] completed step: 3, seconds: 0.409, TFLOP/s/device: 33.235, Tokens/s/device: 5009.503, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0423 15:44:19.963018 136127590459200 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0423 15:44:20.251597 136127590459200 checkpointing.py:798] Waited 0.28855085372924805 seconds for step 5 to finish before starting checkpointing. I0423 15:44:20.254389 136127590459200 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:44:22.178165 2761 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0423 15:44:24.230922 136000121091840 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/0/items/array_metadatas/process_6 I0423 15:45:01.164925 136001737815808 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.118 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.55663990974426 s) (per-host) I0423 15:45:01.165029 136001737815808 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.271326s. I0423 15:45:08.876252 136001737815808 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.982532s. I0423 15:45:08.876539 136000101250816 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:45:08.876664 136000101250816 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:45:08.876714 136000101250816 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:45:08.879225 136000101250816 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 15:45:08.879397 136127590459200 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. W0423 15:45:08.879523 136127590459200 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.625135 seconds. If this number is high, consider checkpointing less frequently. I0423 15:45:08.881634 136127590459200 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0423 15:45:08.884278 136127590459200 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/5. I0423 15:45:09.221648 136127590459200 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:45:09.221745 136127590459200 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 I0423 15:45:09.257001 136127590459200 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.036978s I0423 15:45:09.257179 136127590459200 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.578 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.044611215591430664 s) (per-host) I0423 15:45:09.257230 136127590459200 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.044672s (batch_requests_ready=0.002890s, total_serialization_initiated=0.041713s, others=0.000069s) I0423 15:45:09.257324 136127590459200 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049699s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=0.049678) I0423 15:45:09.257990 136127590459200 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.38 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/5. I0423 15:45:09.258337 136000101250816 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:05:09.258297 I0423 15:45:09.275565 136127590459200 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:45:09.275840 136000121091840 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:45:09.276002 136127590459200 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959060.2543597, 'wait_for_prev_duration_secs': 48.62513494491577, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776959108.8817163, 'checkpointer_blocking_duration_secs': 0.3767237663269043, 'get_old_steps_start_time': 1776959109.2584612, 'get_old_steps_duration_secs': 3.24249267578125e-05, 'checkpoint_manager_blocking_start_time': 1776959060.2519507, 'checkpoint_manager_blocking_duration_secs': 49.02401804924011} I0423 15:45:09.276128 136127590459200 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0423 15:45:09.277225 136127590459200 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.493, Tokens/s/device: 12434.277, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0423 15:45:09.284927 136127590459200 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.280, Tokens/s/device: 13909.453, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0423 15:45:14.430279 135979339851520 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/5/items/array_metadatas/process_6 I0423 15:45:32.683347 136127590459200 metric_logger.py:196] completed step: 6, seconds: 49.320, TFLOP/s/device: 0.275, Tokens/s/device: 41.525, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0423 15:45:32.831085 136127590459200 metric_logger.py:196] completed step: 7, seconds: 23.253, TFLOP/s/device: 0.584, Tokens/s/device: 88.074, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0423 15:45:32.978601 136127590459200 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.696, Tokens/s/device: 13369.193, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0423 15:45:33.126724 136127590459200 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0423 15:45:33.127655 136127590459200 checkpointing.py:798] Waited 0.0009064674377441406 seconds for step 9 to finish before starting checkpointing. I0423 15:45:33.129920 136127590459200 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:45:52.365763 136000101250816 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.604 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.15315818786621 s) (per-host) I0423 15:45:52.365885 136000101250816 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.106936s. I0423 15:45:59.652985 136000101250816 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.394021s. I0423 15:45:59.653300 136000121091840 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:45:59.653421 136000121091840 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:45:59.653473 136000121091840 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:45:59.654955 136000121091840 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0423 15:45:59.655143 136127590459200 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. W0423 15:45:59.655268 136127590459200 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.525354 seconds. If this number is high, consider checkpointing less frequently. I0423 15:45:59.656732 136127590459200 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0423 15:45:59.658891 136127590459200 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9. I0423 15:45:59.981788 136127590459200 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0423 15:45:59.981878 136127590459200 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 I0423 15:46:00.015759 136127590459200 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035571s I0423 15:46:00.015919 136127590459200 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 35.962 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.042894601821899414 s) (per-host) I0423 15:46:00.015969 136127590459200 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.042956s (batch_requests_ready=0.003050s, total_serialization_initiated=0.039840s, others=0.000067s) I0423 15:46:00.016057 136127590459200 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.046905s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.046894) I0423 15:46:00.016842 136127590459200 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9. I0423 15:46:00.017138 136000121091840 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-23 16:06:00.017106 I0423 15:46:00.019108 136127590459200 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 15:46:00.019340 135979339851520 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0423 15:46:00.019444 136127590459200 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776959133.129888, 'wait_for_prev_duration_secs': 26.525354385375977, 'time_between_consecutive_saves_sec': 24.25062394142151, 'checkpointer_blocking_start_time': 1776959159.6567729, 'checkpointer_blocking_duration_secs': 0.3605020046234131, 'get_old_steps_start_time': 1776959160.0172932, 'get_old_steps_duration_secs': 2.4318695068359375e-05, 'checkpoint_manager_blocking_start_time': 1776959133.1280754, 'checkpoint_manager_blocking_duration_secs': 26.891338109970093} I0423 15:46:00.019608 136127590459200 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0423 15:46:00.019651 136127590459200 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 15:46:06.163431 135999040505600 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_20260423_124541/nnx_xpk_feat_nnx_post_train_fixes_20260423_124541_16_zarr3_false/checkpoints/9/items/array_metadatas/process_6 I0423 15:46:42.500835 136000121091840 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.143 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.52777314186096 s) (per-host) I0423 15:46:42.500954 136000121091840 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.483708s. I0423 15:46:48.600738 136000121091840 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.583477s. I0423 15:46:48.600977 135979339851520 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 15:46:48.601035 135979339851520 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0423 15:46:48.601074 135979339851520 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 15:46:48.602670 135979339851520 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 15:46:48.602910 136127590459200 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. I0423 15:46:48.603078 136127590459200 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 15:46:48.604031 136127590459200 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.796, Tokens/s/device: 13836.529, 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: Thu Apr 23 15:46:56 UTC 2026 EXIT_CODE=0