XPK Start: Tue Apr 21 12:58:09 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-21 12:58:33.615377: 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) I0421 12:58:34.154025 136703472932672 max_utils.py:273] Attempting to initialize the jax distributed system... I0421 12:58:43.195070 136703472932672 distributed.py:149] Starting JAX distributed service on [::]:8482 I0421 12:58:43.197403 136703472932672 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-true-sv-ly91m-slice-job-0-0.mt-16-zarr3-true-sv-ly91m:8482 I0421 12:58:44.072564 136703472932672 max_utils.py:284] Jax distributed system initialized! I0421 12:58:50.119677 136703472932672 max_utils.py:800] System Information: Jax Version: 0.9.2 I0421 12:58:50.119803 136703472932672 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0421 12:58:50.119845 136703472932672 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 I0421 12:58:50.119881 136703472932672 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0421 12:58:50.822689 136703472932672 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 12:58:50.823412 136703472932672 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0421 12:58:50.823621 136703472932672 checkpointing.py:688] Setting up checkpoint logger... I0421 12:58:50.823680 136703472932672 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0421 12:58:50.823752 136703472932672 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0421 12:58:50.824145 136703472932672 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 0x7c540a350320>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0421 12:58:53.569935 136703472932672 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0421 12:58:53.570173 136703472932672 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c3fe8146e40>}, handler_registry=None I0421 12:58:53.570446 136703472932672 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c3fe8146e40>` 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`. I0421 12:58:53.570520 136703472932672 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c3fe814bf80>` 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`. I0421 12:58:53.570564 136703472932672 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 0x7c3fe8146e40>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c3fe8146e40>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c3fe814bf80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c3fe814bf80>}). I0421 12:58:53.571225 136703472932672 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0421 12:58:53.571339 136703472932672 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c3fc84d0180> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0421 12:58:54.865384 136703472932672 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints I0421 12:58:55.277419 136703472932672 checkpoint_manager.py:929] [process=7][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/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c3fe812b950> I0421 12:58:55.277588 136703472932672 checkpointing.py:302] Checkpoint manager created! I0421 12:58:56.215850 136703472932672 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0421 12:58:56.215953 136703472932672 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:58:56.593770 136703472932672 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 12:58:56.593861 136703472932672 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:58:56.610308 136703472932672 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0421 12:58:56.610367 136703472932672 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0421 12:58:56.634783 136703472932672 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:58:56.634850 136703472932672 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:58:56.651382 136703472932672 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:58:56.651446 136703472932672 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:58:56.667927 136703472932672 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0421 12:58:56.667990 136703472932672 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:58:56.692859 136703472932672 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0421 12:58:56.692925 136703472932672 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0421 12:58:56.713858 136703472932672 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0421 12:58:56.713921 136703472932672 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0421 12:58:56.935625 136703472932672 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0421 12:58:56.935745 136703472932672 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0421 12:58:58.375528 136703472932672 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0421 12:58:58.375654 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 12:58:58.375709 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0421 12:58:58.375792 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0421 12:58:58.376097 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0421 12:58:58.376234 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0421 12:58:58.376315 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0421 12:58:58.376383 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: P('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0421 12:58:58.376431 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0421 12:58:58.376470 136703472932672 maxtext_utils.py:1827] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: P('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0421 12:58:58.376526 136703472932672 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0421 12:58:58.376584 136703472932672 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0421 12:58:58.870882 136703472932672 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 12:58:58.870985 136703472932672 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0421 12:58:58.886528 136703472932672 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0421 12:58:58.886588 136703472932672 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0421 12:59:09.805814 136703472932672 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. I0421 12:59:09.806626 136703472932672 metric_logger.py:301] number parameters: 1.104 billion I0421 12:59:21.085998 136703472932672 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0421 12:59:21.240037 136703472932672 checkpointing.py:798] Waited 0.1540215015411377 seconds for step 0 to finish before starting checkpointing. I0421 12:59:21.242456 136703472932672 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 12:59:21.244128 136703472932672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0 I0421 12:59:21.245478 136703472932672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/0. I0421 12:59:22.017990 136703472932672 signaling_client.py:364] Using JaxDistributedSignalingClient I0421 12:59:22.018988 136703472932672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 12:59:22.019045 136703472932672 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 I0421 12:59:22.307908 136703472932672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.289995s I0421 12:59:22.308088 136703472932672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.219 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.29558277130126953 s) (per-host) I0421 12:59:22.308140 136703472932672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.295651s (batch_requests_ready=0.002345s, total_serialization_initiated=0.293230s, others=0.000077s) I0421 12:59:22.308237 136703472932672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.299850s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.299832) I0421 12:59:22.309052 136703472932672 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.06 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/0. I0421 12:59:22.309360 136575033562880 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:19:22.309317 I0421 12:59:22.311329 136703472932672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0421 12:59:22.311569 136573433665280 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 12:59:22.311685 136703472932672 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776361.2424347, 'wait_for_prev_duration_secs': 7.271766662597656e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776776361.2441673, 'checkpointer_blocking_duration_secs': 1.0653514862060547, 'get_old_steps_start_time': 1776776362.3095436, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776776361.2405465, 'checkpoint_manager_blocking_duration_secs': 1.071101427078247} I0421 12:59:22.311801 136703472932672 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0421 12:59:22.311850 136703472932672 max_utils.py:750] Memstats: After params initialized: I0421 12:59:22.311898 136703472932672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0421 12:59:22.311930 136703472932672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0421 12:59:22.311959 136703472932672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0421 12:59:22.311987 136703472932672 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0421 12:59:22.621781 136703472932672 metric_logger.py:196] completed step: 0, seconds: 11.279, TFLOP/s/device: 1.205, Tokens/s/device: 181.572, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0421 12:59:22.813716 136703472932672 metric_logger.py:196] completed step: 1, seconds: 1.534, TFLOP/s/device: 8.856, Tokens/s/device: 1334.820, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0421 12:59:23.214478 136703472932672 metric_logger.py:196] completed step: 2, seconds: 0.045, TFLOP/s/device: 304.829, Tokens/s/device: 45947.098, total_weights: 65536, loss: 10.560, lm_loss: 10.560, perplexity: 38547.719 I0421 12:59:23.361932 136703472932672 metric_logger.py:196] completed step: 3, seconds: 0.401, TFLOP/s/device: 33.898, Tokens/s/device: 5109.513, total_weights: 65536, loss: 9.980, lm_loss: 9.980, perplexity: 21590.506 I0421 12:59:23.366297 136703472932672 checkpointing.py:794] Waiting for step 5 to finish before checkpoint... I0421 12:59:23.655908 136703472932672 checkpointing.py:798] Waited 0.28957509994506836 seconds for step 5 to finish before starting checkpointing. I0421 12:59:23.658951 136703472932672 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 12:59:25.440608 2801 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0421 12:59:27.488238 136573442057984 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/0/items/array_metadatas/process_7 I0421 12:59:58.644573 136575033562880 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 43.121 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.63203477859497 s) (per-host) I0421 12:59:58.644699 136575033562880 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.335220s. I0421 13:00:07.889561 136575033562880 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 45.580064s. I0421 13:00:07.889816 136573433665280 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 13:00:07.889873 136573433665280 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 13:00:07.889930 136573433665280 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0421 13:00:07.891623 136573433665280 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0421 13:00:07.891832 136703472932672 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0421 13:00:07.891973 136703472932672 checkpoint_manager.py:1452] Waiting for previous save to complete took 44.233021 seconds. If this number is high, consider checkpointing less frequently. I0421 13:00:07.893811 136703472932672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 5 I0421 13:00:07.896184 136703472932672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/5. I0421 13:00:08.218485 136703472932672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 13:00:08.218579 136703472932672 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 I0421 13:00:08.253154 136703472932672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035754s I0421 13:00:08.253331 136703472932672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 39.268 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.039282798767089844 s) (per-host) I0421 13:00:08.253384 136703472932672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039346s (batch_requests_ready=0.001863s, total_serialization_initiated=0.037411s, others=0.000072s) I0421 13:00:08.253474 136703472932672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.044319s (all_items=0.000015s, per_item={'items': '0.00001454'}, temp_paths=0.044304) I0421 13:00:08.254233 136703472932672 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/5. I0421 13:00:08.254562 136573433665280 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:20:08.254527 I0421 13:00:08.263605 136703472932672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=5] Starting CheckpointManager Save Finalize thread=save_finalize I0421 13:00:08.263863 136573442057984 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 13:00:08.264004 136703472932672 standard_logger.py:34] {'step': 5, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776363.6589208, 'wait_for_prev_duration_secs': 44.23302102088928, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776776407.8938506, 'checkpointer_blocking_duration_secs': 0.3608560562133789, 'get_old_steps_start_time': 1776776408.2547414, 'get_old_steps_duration_secs': 3.147125244140625e-05, 'checkpoint_manager_blocking_start_time': 1776776363.6564965, 'checkpoint_manager_blocking_duration_secs': 44.607471227645874} I0421 13:00:08.264122 136703472932672 checkpointing.py:409] Started an asynchronous checkpoint save for step 5 I0421 13:00:08.265026 136703472932672 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.719, Tokens/s/device: 13372.685, total_weights: 65536, loss: 9.460, lm_loss: 9.460, perplexity: 12840.355 I0421 13:00:08.271314 136703472932672 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.194, Tokens/s/device: 13896.428, total_weights: 65536, loss: 8.959, lm_loss: 8.959, perplexity: 7776.467 I0421 13:00:13.623056 136552864782080 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/5/items/array_metadatas/process_7 I0421 13:00:30.825252 136703472932672 metric_logger.py:196] completed step: 6, seconds: 44.903, TFLOP/s/device: 0.303, Tokens/s/device: 45.609, total_weights: 65536, loss: 8.469, lm_loss: 8.469, perplexity: 4762.740 I0421 13:00:30.972836 136703472932672 metric_logger.py:196] completed step: 7, seconds: 22.408, TFLOP/s/device: 0.606, Tokens/s/device: 91.396, total_weights: 65536, loss: 8.003, lm_loss: 8.003, perplexity: 2989.151 I0421 13:00:31.120067 136703472932672 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.161, Tokens/s/device: 13439.378, total_weights: 65536, loss: 7.572, lm_loss: 7.572, perplexity: 1943.098 I0421 13:00:31.266634 136703472932672 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0421 13:00:31.267351 136703472932672 checkpointing.py:798] Waited 0.0007381439208984375 seconds for step 9 to finish before starting checkpointing. I0421 13:00:31.269466 136703472932672 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=5][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 13:00:50.094066 136573433665280 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.717 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.87999939918518 s) (per-host) I0421 13:00:50.094174 136573433665280 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.839504s. I0421 13:00:57.312697 136573433665280 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 49.058010s. I0421 13:00:57.312989 136573442057984 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 13:00:57.313109 136573442057984 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 13:00:57.313158 136573442057984 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is syncing with other hosts... I0421 13:00:57.314812 136573442057984 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=5] CheckpointManager Save Finalize is done on all hosts. I0421 13:00:57.314977 136703472932672 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=5][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=5. W0421 13:00:57.315113 136703472932672 checkpoint_manager.py:1452] Waiting for previous save to complete took 26.045654 seconds. If this number is high, consider checkpointing less frequently. I0421 13:00:57.316718 136703472932672 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9 I0421 13:00:57.318649 136703472932672 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/9. I0421 13:00:57.624384 136703472932672 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0421 13:00:57.624473 136703472932672 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 I0421 13:00:57.663701 136703472932672 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.040081s I0421 13:00:57.663878 136703472932672 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 35.379 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04360151290893555 s) (per-host) I0421 13:00:57.663931 136703472932672 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.043665s (batch_requests_ready=0.001739s, total_serialization_initiated=0.041857s, others=0.000069s) I0421 13:00:57.664022 136703472932672 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.047606s (all_items=0.000011s, per_item={'items': '0.00001073'}, temp_paths=0.047596) I0421 13:00:57.664663 136703472932672 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/9. I0421 13:00:57.664993 136552696993536 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-21 13:20:57.664959 I0421 13:00:57.666899 136703472932672 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0421 13:00:57.667187 136573442057984 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0421 13:00:57.667318 136703472932672 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776776431.2694364, 'wait_for_prev_duration_secs': 26.04565405845642, 'time_between_consecutive_saves_sec': 23.377772092819214, 'checkpointer_blocking_start_time': 1776776457.3167696, 'checkpointer_blocking_duration_secs': 0.348372220993042, 'get_old_steps_start_time': 1776776457.6651592, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776776431.2675822, 'checkpoint_manager_blocking_duration_secs': 26.399707078933716} I0421 13:00:57.667475 136703472932672 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0421 13:00:57.667520 136703472932672 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0421 13:01:02.846846 136553401661184 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260421_114057/linen_xpk_feat_nnx_post_train_fixes_20260421_114057_16_zarr3_true/checkpoints/9/items/array_metadatas/process_7 I0421 13:01:40.053457 136552696993536 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.226 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.433138847351074 s) (per-host) I0421 13:01:40.053566 136552696993536 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.388459s. I0421 13:01:46.940275 136552696993536 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 49.275153s. I0421 13:01:46.940536 136573442057984 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0421 13:01:46.940652 136573442057984 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0421 13:01:46.940693 136573442057984 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0421 13:01:46.942290 136573442057984 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0421 13:01:46.942465 136703472932672 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0421 13:01:46.942606 136703472932672 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0421 13:01:46.943660 136703472932672 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.191, Tokens/s/device: 13895.957, total_weights: 65536, loss: 7.181, lm_loss: 7.181, perplexity: 1314.495 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Tue Apr 21 13:01:59 UTC 2026 EXIT_CODE=0