XPK Start: Wed Apr 22 15:10:08 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-22 15:10:32.846666: 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) I0422 15:10:33.029862 140397929723712 max_utils.py:273] Attempting to initialize the jax distributed system... I0422 15:10:42.073024 140397929723712 distributed.py:149] Starting JAX distributed service on [::]:8482 I0422 15:10:42.075970 140397929723712 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-true-sv-j1uqq-slice-job-0-0.mt-15-ocdbt-true-sv-j1uqq:8482 I0422 15:10:43.826430 140397929723712 max_utils.py:284] Jax distributed system initialized! I0422 15:10:49.877495 140397929723712 max_utils.py:800] System Information: Jax Version: 0.9.2 I0422 15:10:49.877597 140397929723712 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0422 15:10:49.877637 140397929723712 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 I0422 15:10:49.877672 140397929723712 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0422 15:10:50.579008 140397929723712 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0422 15:10:50.690873 140397929723712 checkpointing.py:688] Setting up checkpoint logger... I0422 15:10:50.690983 140397929723712 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0422 15:10:50.691028 140397929723712 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0422 15:10:50.691241 140397929723712 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 0x7fb05a320230>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0422 15:10:53.657289 140397929723712 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0422 15:10:53.657530 140397929723712 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 0x7f9c144e88f0>}, handler_registry=None I0422 15:10:53.657769 140397929723712 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9c144e88f0>` 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`. I0422 15:10:53.657817 140397929723712 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9bf421a8d0>` 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`. I0422 15:10:53.657852 140397929723712 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 0x7f9c144e88f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7f9c144e88f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9bf421a8d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7f9bf421a8d0>}). I0422 15:10:53.658191 140397929723712 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0422 15:10:53.658261 140397929723712 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7f9bf42e1da0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0422 15:10:54.705434 140397929723712 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints I0422 15:10:55.013830 140397929723712 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7f9bf420a840> I0422 15:10:55.014002 140397929723712 checkpointing.py:302] Checkpoint manager created! I0422 15:10:55.980281 140397929723712 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0422 15:10:55.980399 140397929723712 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0422 15:10:57.385765 140397929723712 maxtext_utils.py:1836] decoder/decoder_norm/scale/value Shape: float32[2048] Physical: (None,) I0422 15:10:57.385879 140397929723712 maxtext_utils.py:1836] decoder/layers/mlp/wi_0/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:10:57.385930 140397929723712 maxtext_utils.py:1836] decoder/layers/mlp/wi_1/kernel/value Shape: float32[2048,16,7168] Physical: ('fsdp', None, None) I0422 15:10:57.385972 140397929723712 maxtext_utils.py:1836] decoder/layers/mlp/wo/kernel/value Shape: float32[7168,16,2048] Physical: (None, None, 'fsdp') I0422 15:10:57.386008 140397929723712 maxtext_utils.py:1836] decoder/layers/post_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:10:57.386041 140397929723712 maxtext_utils.py:1836] decoder/layers/pre_self_attention_layer_norm/scale/value Shape: float32[2048,16] Physical: (None, None) I0422 15:10:57.386077 140397929723712 maxtext_utils.py:1836] decoder/layers/self_attention/key/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:10:57.386124 140397929723712 maxtext_utils.py:1836] decoder/layers/self_attention/out/kernel/value Shape: float32[16,16,128,2048] Physical: (None, None, None, 'fsdp') I0422 15:10:57.386159 140397929723712 maxtext_utils.py:1836] decoder/layers/self_attention/query/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:10:57.386192 140397929723712 maxtext_utils.py:1836] decoder/layers/self_attention/value/kernel/value Shape: float32[2048,16,16,128] Physical: ('fsdp', None, None, None) I0422 15:10:57.386224 140397929723712 maxtext_utils.py:1836] decoder/logits_dense/kernel/value Shape: float32[2048,32000] Physical: ('fsdp', None) I0422 15:10:57.386254 140397929723712 maxtext_utils.py:1836] token_embedder/embedding/value Shape: float32[32000,2048] Physical: (None, 'fsdp') I0422 15:10:57.786288 140397929723712 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:10:57.786377 140397929723712 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:10:57.791600 140397929723712 nnx_decoders.py:465] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0422 15:10:57.791655 140397929723712 nnx_decoders.py:465] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:10:57.808169 140397929723712 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:10:57.808228 140397929723712 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:10:57.824066 140397929723712 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0422 15:10:57.824135 140397929723712 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0422 15:10:57.849001 140397929723712 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:10:57.849068 140397929723712 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:10:57.864995 140397929723712 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:10:57.865053 140397929723712 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:10:57.880847 140397929723712 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0422 15:10:57.880908 140397929723712 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:10:57.911677 140397929723712 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0422 15:10:57.911752 140397929723712 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0422 15:10:57.931514 140397929723712 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0422 15:10:57.931584 140397929723712 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0422 15:11:09.523419 140397929723712 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. I0422 15:11:09.525808 140397929723712 metric_logger.py:301] number parameters: 1.104 billion I0422 15:11:20.872281 140397929723712 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0422 15:11:21.024420 140397929723712 checkpointing.py:798] Waited 0.15212392807006836 seconds for step 0 to finish before starting checkpointing. I0422 15:11:21.026787 140397929723712 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:11:21.028424 140397929723712 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0 I0422 15:11:21.029801 140397929723712 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/0. I0422 15:11:21.366091 140397929723712 signaling_client.py:364] Using JaxDistributedSignalingClient I0422 15:11:21.366986 140397929723712 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:11:21.367047 140397929723712 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 I0422 15:11:21.641524 140397929723712 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.276197s I0422 15:11:21.641692 140397929723712 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.402 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2855410575866699 s) (per-host) I0422 15:11:21.641760 140397929723712 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285620s (batch_requests_ready=0.003260s, total_serialization_initiated=0.282273s, others=0.000088s) I0422 15:11:21.641865 140397929723712 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289868s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.289850) I0422 15:11:21.642663 140397929723712 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/0. I0422 15:11:21.642937 140269864621824 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:31:21.642905 I0422 15:11:21.651202 140397929723712 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:11:21.651502 140268263175936 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:11:21.651640 140397929723712 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776870681.0267684, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776870681.0284615, 'checkpointer_blocking_duration_secs': 0.6146233081817627, 'get_old_steps_start_time': 1776870681.6431243, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776870681.024859, 'checkpoint_manager_blocking_duration_secs': 0.6267435550689697} I0422 15:11:21.651803 140397929723712 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0422 15:11:21.651877 140397929723712 max_utils.py:750] Memstats: After params initialized: I0422 15:11:21.651926 140397929723712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0422 15:11:21.651958 140397929723712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0422 15:11:21.651985 140397929723712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0422 15:11:21.652010 140397929723712 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0422 15:11:21.964026 140397929723712 metric_logger.py:196] completed step: 0, seconds: 11.345, TFLOP/s/device: 1.198, Tokens/s/device: 180.514, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0422 15:11:22.150005 140397929723712 metric_logger.py:196] completed step: 1, seconds: 1.090, TFLOP/s/device: 12.468, Tokens/s/device: 1879.363, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52786.469 I0422 15:11:22.592640 140397929723712 metric_logger.py:196] completed step: 2, seconds: 0.040, TFLOP/s/device: 340.410, Tokens/s/device: 51310.317, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38565.637 I0422 15:11:22.740228 140397929723712 metric_logger.py:196] completed step: 3, seconds: 0.432, TFLOP/s/device: 31.434, Tokens/s/device: 4738.021, total_weights: 65536, loss: 9.979, lm_loss: 9.979, perplexity: 21564.932 I0422 15:11:22.746014 140397929723712 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0422 15:11:23.035053 140397929723712 checkpointing.py:798] Waited 0.28901147842407227 seconds for step 5 to finish before starting checkpointing. I0422 15:11:23.037789 140397929723712 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:11:25.407721 2808 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0422 15:11:27.980202 140269280532224 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/0/items/array_metadatas/process_6 I0422 15:11:59.248188 140269864621824 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 41.687 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.89200234413147 s) (per-host) I0422 15:11:59.248315 140269864621824 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.605271s. I0422 15:12:08.100690 140269864621824 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 46.457628s. I0422 15:12:08.100962 140268263175936 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:12:08.101082 140268263175936 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:12:08.101153 140268263175936 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:12:08.103873 140268263175936 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0422 15:12:08.104038 140397929723712 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. W0422 15:12:08.104186 140397929723712 checkpoint_manager.py:1452] Waiting for previous save to complete took 45.066397 seconds. If this number is high, consider checkpointing less frequently. I0422 15:12:08.106250 140397929723712 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 5 I0422 15:12:08.108784 140397929723712 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/5. I0422 15:12:08.412631 140397929723712 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:12:08.412724 140397929723712 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 I0422 15:12:08.457477 140397929723712 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046518s I0422 15:12:08.457648 140397929723712 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 28.730 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.05369305610656738 s) (per-host) I0422 15:12:08.457702 140397929723712 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.053757s (batch_requests_ready=0.002553s, total_serialization_initiated=0.051132s, others=0.000072s) I0422 15:12:08.457795 140397929723712 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.058635s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.058619) I0422 15:12:08.458501 140397929723712 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/5. I0422 15:12:08.458842 140268263175936 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:32:08.458806 I0422 15:12:08.466480 140397929723712 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:12:08.466737 140244229809920 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:12:08.466901 140397929723712 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776870683.0377593, 'wait_for_prev_duration_secs': 45.066396713256836, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776870728.106289, 'checkpointer_blocking_duration_secs': 0.35269665718078613, 'get_old_steps_start_time': 1776870728.4590082, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776870683.0354345, 'checkpoint_manager_blocking_duration_secs': 45.43143177032471} I0422 15:12:08.467018 140397929723712 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0422 15:12:08.468114 140397929723712 metric_logger.py:196] completed step: 4, seconds: 0.165, TFLOP/s/device: 82.520, Tokens/s/device: 12438.355, total_weights: 65536, loss: 9.457, lm_loss: 9.457, perplexity: 12802.436 I0422 15:12:08.475911 140397929723712 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.354, Tokens/s/device: 13920.609, total_weights: 65536, loss: 8.955, lm_loss: 8.955, perplexity: 7748.778 I0422 15:12:14.307989 140269280532224 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/5/items/array_metadatas/process_6 I0422 15:12:31.436727 140397929723712 metric_logger.py:196] completed step: 6, seconds: 45.728, TFLOP/s/device: 0.297, Tokens/s/device: 44.787, total_weights: 65536, loss: 8.465, lm_loss: 8.465, perplexity: 4743.984 I0422 15:12:31.584118 140397929723712 metric_logger.py:196] completed step: 7, seconds: 22.816, TFLOP/s/device: 0.596, Tokens/s/device: 89.763, total_weights: 65536, loss: 7.998, lm_loss: 7.998, perplexity: 2976.266 I0422 15:12:31.731642 140397929723712 metric_logger.py:196] completed step: 8, seconds: 0.153, TFLOP/s/device: 88.732, Tokens/s/device: 13374.607, total_weights: 65536, loss: 7.567, lm_loss: 7.567, perplexity: 1933.479 I0422 15:12:31.879933 140397929723712 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0422 15:12:31.880634 140397929723712 checkpointing.py:798] Waited 0.0007128715515136719 seconds for step 9 to finish before starting checkpointing. I0422 15:12:31.882851 140397929723712 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:12:50.483254 140268263175936 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.539 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.07927870750427 s) (per-host) I0422 15:12:50.483366 140268263175936 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.024410s. I0422 15:12:59.362841 140268263175936 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.903870s. I0422 15:12:59.363329 140244229809920 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:12:59.363446 140244229809920 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:12:59.363495 140244229809920 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:12:59.365847 140244229809920 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0422 15:12:59.366026 140397929723712 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. W0422 15:12:59.366171 140397929723712 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.483324 seconds. If this number is high, consider checkpointing less frequently. I0422 15:12:59.368345 140397929723712 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9 I0422 15:12:59.370471 140397929723712 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9. I0422 15:12:59.686766 140397929723712 jax_array_handlers.py:360] Scheduling D2H of 69 prioritized jax.Array. I0422 15:12:59.686861 140397929723712 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 I0422 15:12:59.728119 140397929723712 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.043005s I0422 15:12:59.728273 140397929723712 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 30.969 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04981088638305664 s) (per-host) I0422 15:12:59.728321 140397929723712 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.049870s (batch_requests_ready=0.002614s, total_serialization_initiated=0.047192s, others=0.000064s) I0422 15:12:59.728414 140397929723712 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053959s (all_items=0.000011s, per_item={'items': '0.00001144'}, temp_paths=0.053947) I0422 15:12:59.729078 140397929723712 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9. I0422 15:12:59.729416 140269280532224 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-22 15:32:59.729379 I0422 15:12:59.731330 140397929723712 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0422 15:12:59.731611 140244229809920 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0422 15:12:59.731745 140397929723712 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/nnx_ckpt_xpk_feat_nnx_post_train_fixes_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776870751.8828232, 'wait_for_prev_duration_secs': 27.483323574066162, 'time_between_consecutive_saves_sec': 23.77890968322754, 'checkpointer_blocking_start_time': 1776870779.368385, 'checkpointer_blocking_duration_secs': 0.3611764907836914, 'get_old_steps_start_time': 1776870779.7295783, 'get_old_steps_duration_secs': 2.47955322265625e-05, 'checkpoint_manager_blocking_start_time': 1776870751.8808684, 'checkpoint_manager_blocking_duration_secs': 27.850844383239746} I0422 15:12:59.731916 140397929723712 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0422 15:12:59.731958 140397929723712 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0422 15:13:04.818220 140267709519616 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_20260422_123906/nnx_xpk_feat_nnx_post_train_fixes_20260422_123906_15_ocdbt_true/checkpoints/9/items/array_metadatas/process_6 I0422 15:13:40.503371 140269280532224 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.692 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.824870109558105 s) (per-host) I0422 15:13:40.503488 140269280532224 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.773962s. I0422 15:13:50.011638 140269280532224 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.282095s. I0422 15:13:50.011903 140244229809920 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0422 15:13:50.012020 140244229809920 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0422 15:13:50.012060 140244229809920 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0422 15:13:50.014321 140244229809920 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0422 15:13:50.014491 140397929723712 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. I0422 15:13:50.014660 140397929723712 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0422 15:13:50.015690 140397929723712 metric_logger.py:196] completed step: 9, seconds: 0.148, TFLOP/s/device: 91.965, Tokens/s/device: 13861.909, 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: Wed Apr 22 15:13:58 UTC 2026 EXIT_CODE=0