XPK Start: Mon Apr 20 17:10:57 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-20 17:11:22.214512: 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) I0420 17:11:22.393760 132681154398016 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 17:11:31.434228 132681154398016 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 17:11:31.436611 132681154398016 distributed.py:172] Connecting to JAX distributed service on mt-15-ocdbt-false-rs-sf0i6-slice-job-0-0.mt-15-ocdbt-false-rs-sf0i6:8482 I0420 17:11:32.283392 132681154398016 max_utils.py:284] Jax distributed system initialized! I0420 17:11:38.522534 132681154398016 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 17:11:38.522636 132681154398016 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 17:11:38.522675 132681154398016 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 I0420 17:11:38.522707 132681154398016 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 17:11:39.218473 132681154398016 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 17:11:39.219069 132681154398016 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 17:11:39.219249 132681154398016 checkpointing.py:688] Setting up checkpoint logger... I0420 17:11:39.219300 132681154398016 checkpointing.py:234] Creating checkpoint manager with ocdbt=False and zarr3=True I0420 17:11:39.219341 132681154398016 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 17:11:39.219696 132681154398016 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=False, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x78aba1118c20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 17:11:40.851759 132681154398016 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 17:11:40.851997 132681154398016 checkpoint_manager.py:708] [process=2][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78975055da60>}, handler_registry=None I0420 17:11:40.852233 132681154398016 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78975055da60>` 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`. I0420 17:11:40.852281 132681154398016 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x789750560e60>` 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`. I0420 17:11:40.852316 132681154398016 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 0x78975055da60>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78975055da60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x789750560e60>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x789750560e60>}). I0420 17:11:40.852635 132681154398016 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 17:11:40.852704 132681154398016 async_checkpointer.py:192] [process=2][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x789660193740> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 17:11:42.846166 132559920944896 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776704923270017398, 'commit_timestamp_nsecs': 1776704938864330270, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/0/_CHECKPOINT_METADATA I0420 17:11:43.109221 132559920944896 checkpoint_manager.py:1871] Missing metrics for step 0 I0420 17:11:43.518576 132681154398016 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints I0420 17:11:44.405283 132681154398016 checkpoint_manager.py:929] [process=2][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_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78975055fa40> I0420 17:11:44.405456 132681154398016 checkpointing.py:302] Checkpoint manager created! I0420 17:11:45.342062 132681154398016 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 17:11:45.342167 132681154398016 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:11:45.719572 132681154398016 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:11:45.719662 132681154398016 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:11:45.735833 132681154398016 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 17:11:45.735891 132681154398016 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 17:11:45.759547 132681154398016 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:11:45.759629 132681154398016 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:11:45.776046 132681154398016 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:11:45.776108 132681154398016 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:11:45.792277 132681154398016 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 17:11:45.792337 132681154398016 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:11:45.816982 132681154398016 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 17:11:45.817058 132681154398016 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 17:11:45.838219 132681154398016 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 17:11:45.838283 132681154398016 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 17:11:46.055425 132681154398016 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 17:11:46.055545 132681154398016 checkpointing.py:582] restoring from this run's directory step 9 I0420 17:11:46.302424 132681154398016 event_tracking.py:70] [process=2] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9. I0420 17:11:46.562174 132681154398016 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9. I0420 17:11:46.562606 132681154398016 event_tracking.py:125] [process=2] [sync] Finished blocking load in 0.26 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9. I0420 17:11:49.238591 1532 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 17:11:58.260395 132681154398016 jax_array_handlers.py:843] [process=2] /jax/orbax/read/worker/io/requested throughput: 155.818 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.14176058769226 s) (per-host) I0420 17:11:58.268537 132681154398016 base_pytree_checkpoint_handler.py:130] [process=2] /jax/checkpoint/read/gbytes_per_sec: 150.086 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.544566631317139 s) (per-host) I0420 17:12:00.393957 132681154398016 event_tracking.py:138] [process=2] [sync] Finished load in 14.09 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/9 I0420 17:12:00.394111 132681154398016 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'checkpointer_start_time': 1776705106.3023255, 'checkpointer_duration_secs': 14.091752767562866, 'checkpoint_manager_start_time': 1776705106.0558043, 'checkpoint_manager_duration_secs': 14.338276147842407} fsdp: 32 I0420 17:12:00.808315 132681154398016 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0420 17:12:00.808436 132681154398016 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) I0420 17:12:00.808488 132681154398016 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) I0420 17:12:00.808559 132681154398016 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') I0420 17:12:00.808611 132681154398016 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) I0420 17:12:00.808648 132681154398016 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) I0420 17:12:00.808699 132681154398016 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) I0420 17:12:00.808750 132681154398016 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') I0420 17:12:00.808789 132681154398016 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) I0420 17:12:00.808824 132681154398016 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) I0420 17:12:00.808870 132681154398016 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0420 17:12:00.808914 132681154398016 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0420 17:12:01.295685 132681154398016 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 17:12:01.295781 132681154398016 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 17:12:01.311019 132681154398016 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 17:12:01.311078 132681154398016 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 17:12:12.185966 132681154398016 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. I0420 17:12:12.186760 132681154398016 metric_logger.py:301] number parameters: 1.104 billion I0420 17:12:23.640454 132681154398016 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0420 17:12:23.794902 132681154398016 checkpointing.py:798] Waited 0.15442943572998047 seconds for step 10 to finish before starting checkpointing. I0420 17:12:23.797564 132681154398016 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:12:23.799201 132681154398016 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 10 I0420 17:12:23.800607 132681154398016 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/10. I0420 17:12:24.133874 132681154398016 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 17:12:24.136013 132681154398016 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 17:12:24.136081 132681154398016 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 I0420 17:12:24.415167 132681154398016 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.281366s I0420 17:12:24.415341 132681154398016 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 5.410 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28514885902404785 s) (per-host) I0420 17:12:24.415394 132681154398016 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.285213s (batch_requests_ready=0.002089s, total_serialization_initiated=0.283051s, others=0.000073s) I0420 17:12:24.415482 132681154398016 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.289482s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.289464) I0420 17:12:24.416259 132681154398016 event_tracking.py:125] [process=2] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/10. I0420 17:12:24.416620 132555307218688 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 17:32:24.416581 I0420 17:12:24.418658 132681154398016 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:12:24.418890 132554761955072 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:12:24.419046 132681154398016 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776705143.797545, 'wait_for_prev_duration_secs': 5.9604644775390625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776705143.799239, 'checkpointer_blocking_duration_secs': 0.6175463199615479, 'get_old_steps_start_time': 1776705144.4168088, 'get_old_steps_duration_secs': 3.2901763916015625e-05, 'checkpoint_manager_blocking_start_time': 1776705143.7954235, 'checkpoint_manager_blocking_duration_secs': 0.6235842704772949} I0420 17:12:24.419158 132681154398016 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0420 17:12:24.419209 132681154398016 max_utils.py:750] Memstats: After params initialized: I0420 17:12:24.419255 132681154398016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_8(process=2,(0,2,0,0)) I0420 17:12:24.419289 132681154398016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_9(process=2,(1,2,0,0)) I0420 17:12:24.419316 132681154398016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_12(process=2,(0,3,0,0)) I0420 17:12:24.419341 132681154398016 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_13(process=2,(1,3,0,0)) I0420 17:12:24.735143 132681154398016 metric_logger.py:196] completed step: 10, seconds: 11.454, TFLOP/s/device: 1.186, Tokens/s/device: 178.809, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0420 17:12:24.909832 132681154398016 metric_logger.py:196] completed step: 11, seconds: 1.093, TFLOP/s/device: 12.429, Tokens/s/device: 1873.471, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0420 17:12:25.057215 132681154398016 metric_logger.py:196] completed step: 12, seconds: 0.028, TFLOP/s/device: 492.162, Tokens/s/device: 74184.084, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0420 17:12:25.204633 132681154398016 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.786, Tokens/s/device: 13382.822, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0420 17:12:25.209231 132681154398016 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0420 17:12:25.498816 132681154398016 checkpointing.py:798] Waited 0.2895491123199463 seconds for step 15 to finish before starting checkpointing. I0420 17:12:25.501818 132681154398016 checkpoint_manager.py:2020] [process=2][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:12:29.721120 132554770347776 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/10/items/array_metadatas/process_2 I0420 17:12:34.768429 132555307218688 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 148.483 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.638203382492065 s) (per-host) I0420 17:12:34.768573 132555307218688 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.351831s. I0420 17:12:41.313124 132555307218688 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 16.896366s. I0420 17:12:41.313414 132554761955072 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:12:41.313550 132554761955072 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:12:41.313614 132554761955072 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:12:41.315889 132554761955072 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0420 17:12:41.316071 132681154398016 checkpoint_manager.py:2032] [process=2][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0420 17:12:41.316206 132681154398016 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.814389 seconds. If this number is high, consider checkpointing less frequently. I0420 17:12:41.318201 132681154398016 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 15 I0420 17:12:41.320578 132681154398016 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/15. I0420 17:12:42.053285 132681154398016 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 17:12:42.053379 132681154398016 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 I0420 17:12:42.079178 132681154398016 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026985s I0420 17:12:42.079303 132681154398016 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 50.872 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030322790145874023 s) (per-host) I0420 17:12:42.079350 132681154398016 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030378s (batch_requests_ready=0.001751s, total_serialization_initiated=0.028565s, others=0.000062s) I0420 17:12:42.079426 132681154398016 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035468s (all_items=0.000013s, per_item={'items': '0.00001287'}, temp_paths=0.035455) I0420 17:12:42.080190 132681154398016 event_tracking.py:125] [process=2] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/15. I0420 17:12:42.080443 132554761955072 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 17:32:42.080421 I0420 17:12:42.092051 132681154398016 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:12:42.092297 132554753562368 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:12:42.092431 132681154398016 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776705145.5017872, 'wait_for_prev_duration_secs': 15.814388751983643, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776705161.3182395, 'checkpointer_blocking_duration_secs': 0.7623195648193359, 'get_old_steps_start_time': 1776705162.0805857, 'get_old_steps_duration_secs': 3.170967102050781e-05, 'checkpoint_manager_blocking_start_time': 1776705145.499434, 'checkpoint_manager_blocking_duration_secs': 16.59296202659607} I0420 17:12:42.092559 132681154398016 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0420 17:12:42.093377 132681154398016 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.179, Tokens/s/device: 13894.166, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0420 17:12:42.099531 132681154398016 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.131, Tokens/s/device: 13887.005, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0420 17:12:47.278496 132554770347776 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/15/items/array_metadatas/process_2 I0420 17:12:52.290635 132554761955072 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 154.233 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 10.241614818572998 s) (per-host) I0420 17:12:52.290757 132554761955072 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 10.210243s. I0420 17:12:57.487387 132554761955072 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 15.406857s. I0420 17:12:57.487632 132554753562368 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:12:57.487686 132554753562368 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:12:57.487731 132554753562368 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:12:57.490401 132554753562368 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0420 17:13:03.325629 132681154398016 metric_logger.py:196] completed step: 16, seconds: 16.889, TFLOP/s/device: 0.805, Tokens/s/device: 121.265, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0420 17:13:03.472823 132681154398016 metric_logger.py:196] completed step: 17, seconds: 21.080, TFLOP/s/device: 0.645, Tokens/s/device: 97.152, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0420 17:13:03.620310 132681154398016 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.252, Tokens/s/device: 13452.974, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0420 17:13:03.767202 132681154398016 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0420 17:13:03.767911 132681154398016 checkpointing.py:798] Waited 0.0007290840148925781 seconds for step 19 to finish before starting checkpointing. I0420 17:13:03.769879 132681154398016 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:13:03.771588 132681154398016 checkpoint_manager.py:1512] [process=2] Saving checkpoint at step 19 I0420 17:13:03.772929 132681154398016 event_tracking.py:70] [process=2] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/19. I0420 17:13:04.501735 132681154398016 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 17:13:04.501824 132681154398016 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 I0420 17:13:04.527455 132681154398016 base_pytree_checkpoint_handler.py:154] [process=2][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026722s I0420 17:13:04.527611 132681154398016 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/blocking_gbytes_per_sec: 51.205 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03012561798095703 s) (per-host) I0420 17:13:04.527662 132681154398016 base_pytree_checkpoint_handler.py:768] [process=2][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030185s (batch_requests_ready=0.001619s, total_serialization_initiated=0.028502s, others=0.000065s) I0420 17:13:04.527745 132681154398016 composite_checkpoint_handler.py:715] [process=2][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034428s (all_items=0.000011s, per_item={'items': '0.00001121'}, temp_paths=0.034417) I0420 17:13:04.528444 132681154398016 event_tracking.py:125] [process=2] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/19. I0420 17:13:04.528760 132554753562368 async_checkpointer.py:76] [process=2][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 17:33:04.528727 I0420 17:13:04.532136 132681154398016 checkpoint_manager.py:1560] [process=2][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0420 17:13:04.532417 132551012251392 async_checkpointer.py:280] [process=2][thread=save_finalize] Waiting for background save thread=async_save. I0420 17:13:04.532584 132681154398016 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776705183.7698507, 'wait_for_prev_duration_secs': 6.890296936035156e-05, 'time_between_consecutive_saves_sec': 6.279408931732178, 'checkpointer_blocking_start_time': 1776705183.7716258, 'checkpointer_blocking_duration_secs': 0.7572753429412842, 'get_old_steps_start_time': 1776705184.528918, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1776705183.7681499, 'checkpoint_manager_blocking_duration_secs': 0.7644026279449463} I0420 17:13:04.532700 132681154398016 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0420 17:13:04.532748 132681154398016 checkpoint_manager.py:2020] [process=2][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 17:13:11.063615 132554770347776 array_metadata_store.py:203] [process=2][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_15_ocdbt_false/checkpoints/19/items/array_metadatas/process_2 I0420 17:13:15.754536 132554753562368 base_pytree_checkpoint_handler.py:130] [process=2] /jax/orbax/write/gbytes_per_sec: 140.321 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 11.257009029388428 s) (per-host) I0420 17:13:15.754638 132554753562368 async_checkpointer.py:90] [process=2][thread=async_save] 3 Handler Commit operations completed. Time taken: 11.225767s. I0420 17:13:21.229149 132554753562368 async_checkpointer.py:160] [process=2][thread=async_save] Background save thread done. Time taken: 16.700261s. I0420 17:13:21.229416 132551012251392 async_checkpointer.py:288] [process=2][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 17:13:21.229553 132551012251392 async_checkpointer.py:298] [process=2][thread=save_finalize] No errors found in background save thread=async_save. I0420 17:13:21.229594 132551012251392 checkpoint_manager.py:2137] [process=2][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0420 17:13:21.231915 132551012251392 checkpoint_manager.py:2146] [process=2][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0420 17:13:21.232107 132681154398016 checkpoint_manager.py:2032] [process=2][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0420 17:13:21.232256 132681154398016 checkpoint_manager.py:2009] [process=2][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 17:13:21.233074 132681154398016 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.234, Tokens/s/device: 13902.560, total_weights: 65536, loss: 5.391, lm_loss: 5.391, perplexity: 219.349 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 17:13:33 UTC 2026 EXIT_CODE=0