XPK Start: Sat Apr 25 13:35:22 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. `rope_parameters`'s factor field must be a float >= 1, got 40 `rope_parameters`'s beta_fast field must be a float, got 32 `rope_parameters`'s beta_slow field must be a float, got 1 DeepseekV32Config got `key=rope_scaling` in kwargs but hasn't set it as attribute. For RoPE standardization you need to set `self.rope_parameters` in model's config. 2026-04-25 13:35:47.249843: 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) I0425 13:35:47.462414 138079697409856 max_utils.py:273] Attempting to initialize the jax distributed system... I0425 13:35:56.504604 138079697409856 distributed.py:149] Starting JAX distributed service on [::]:8482 I0425 13:35:56.507011 138079697409856 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-awit9-slice-job-0-0.mt-16-zarr3-false-rs-awit9:8482 I0425 13:35:57.326679 138079697409856 max_utils.py:284] Jax distributed system initialized! I0425 13:36:03.380860 138079697409856 max_utils.py:800] System Information: Jax Version: 0.9.2 I0425 13:36:03.380967 138079697409856 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0425 13:36:03.381007 138079697409856 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 I0425 13:36:03.381044 138079697409856 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0425 13:36:04.076736 138079697409856 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 13:36:04.077330 138079697409856 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0425 13:36:04.077519 138079697409856 checkpointing.py:688] Setting up checkpoint logger... I0425 13:36:04.077573 138079697409856 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0425 13:36:04.077618 138079697409856 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0425 13:36:04.077956 138079697409856 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=False, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7d949151cf50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0425 13:36:06.170423 138079697409856 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0425 13:36:06.170663 138079697409856 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 0x7d80d83da0c0>}, handler_registry=None I0425 13:36:06.170899 138079697409856 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d80d83da0c0>` 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`. I0425 13:36:06.170947 138079697409856 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d8038751b20>` 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`. I0425 13:36:06.170984 138079697409856 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 0x7d80d83da0c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7d80d83da0c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d8038751b20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d8038751b20>}). I0425 13:36:06.171353 138079697409856 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0425 13:36:06.171421 138079697409856 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d80384e9d00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0425 13:36:07.646599 137961974777600 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777123950689722948, 'commit_timestamp_nsecs': 1777123997083886686, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/5/_CHECKPOINT_METADATA I0425 13:36:07.890492 137961974777600 checkpoint_manager.py:1871] Missing metrics for step 5 I0425 13:36:08.199414 138079697409856 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints I0425 13:36:08.660014 138079697409856 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/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7d94e67d3380> I0425 13:36:08.660226 138079697409856 checkpointing.py:302] Checkpoint manager created! I0425 13:36:09.596505 138079697409856 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0425 13:36:09.596615 138079697409856 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 13:36:09.977547 138079697409856 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 13:36:09.977641 138079697409856 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 13:36:09.994202 138079697409856 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0425 13:36:09.994266 138079697409856 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0425 13:36:10.018451 138079697409856 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 13:36:10.018520 138079697409856 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 13:36:10.035150 138079697409856 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 13:36:10.035211 138079697409856 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 13:36:10.051946 138079697409856 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0425 13:36:10.052013 138079697409856 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 13:36:10.077000 138079697409856 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0425 13:36:10.077070 138079697409856 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0425 13:36:10.098173 138079697409856 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0425 13:36:10.098238 138079697409856 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0425 13:36:10.320034 138079697409856 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0425 13:36:10.320165 138079697409856 checkpointing.py:582] restoring from this run's directory step 9 I0425 13:36:10.568009 138079697409856 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9. I0425 13:36:10.812490 138079697409856 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9. I0425 13:36:10.812937 138079697409856 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.24 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9. I0425 13:36:12.282784 1414 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0425 13:36:18.441549 138079697409856 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 256.344 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.1646504402160645 s) (per-host) I0425 13:36:18.446205 138079697409856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 240.907 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.569323778152466 s) (per-host) I0425 13:36:18.592348 138079697409856 event_tracking.py:138] [process=6] [sync] Finished load in 8.02 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/9 I0425 13:36:18.592508 138079697409856 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1777124170.567911, 'checkpointer_duration_secs': 8.024565935134888, 'checkpoint_manager_start_time': 1777124170.3204343, 'checkpoint_manager_duration_secs': 8.27204442024231} fsdp: 32 I0425 13:36:19.013733 138079697409856 maxtext_utils.py:1880] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0425 13:36:19.013865 138079697409856 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 13:36:19.013921 138079697409856 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0425 13:36:19.014000 138079697409856 maxtext_utils.py:1880] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0425 13:36:19.014058 138079697409856 maxtext_utils.py:1880] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 13:36:19.014150 138079697409856 maxtext_utils.py:1880] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0425 13:36:19.014211 138079697409856 maxtext_utils.py:1880] 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) I0425 13:36:19.014496 138079697409856 maxtext_utils.py:1880] 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') I0425 13:36:19.014670 138079697409856 maxtext_utils.py:1880] 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) I0425 13:36:19.014737 138079697409856 maxtext_utils.py:1880] 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) I0425 13:36:19.014809 138079697409856 maxtext_utils.py:1880] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0425 13:36:19.014871 138079697409856 maxtext_utils.py:1880] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0425 13:36:19.508439 138079697409856 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 13:36:19.508535 138079697409856 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0425 13:36:19.524279 138079697409856 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0425 13:36:19.524342 138079697409856 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0425 13:36:30.660109 138079697409856 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. I0425 13:36:30.660940 138079697409856 metric_logger.py:301] number parameters: 1.104 billion I0425 13:36:42.030927 138079697409856 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0425 13:36:42.229626 138079697409856 checkpointing.py:798] Waited 0.19867587089538574 seconds for step 10 to finish before starting checkpointing. I0425 13:36:42.231978 138079697409856 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 13:36:42.233954 138079697409856 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0425 13:36:42.235356 138079697409856 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10. I0425 13:36:42.990032 138079697409856 signaling_client.py:364] Using JaxDistributedSignalingClient I0425 13:36:42.991075 138079697409856 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 13:36:42.991151 138079697409856 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 I0425 13:36:43.269273 138079697409856 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.279315s I0425 13:36:43.269454 138079697409856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.448 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2831394672393799 s) (per-host) I0425 13:36:43.269508 138079697409856 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283203s (batch_requests_ready=0.002138s, total_serialization_initiated=0.280992s, others=0.000073s) I0425 13:36:43.269608 138079697409856 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287547s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=0.287529) I0425 13:36:43.270333 138079697409856 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.04 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10. I0425 13:36:43.270677 137961370797824 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 13:56:43.270639 I0425 13:36:43.276185 138079697409856 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0425 13:36:43.276438 137959214921472 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 13:36:43.276571 138079697409856 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777124202.2319593, 'wait_for_prev_duration_secs': 6.031990051269531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777124202.2339933, 'checkpointer_blocking_duration_secs': 1.036834478378296, 'get_old_steps_start_time': 1777124203.27085, 'get_old_steps_duration_secs': 3.361701965332031e-05, 'checkpoint_manager_blocking_start_time': 1777124202.2301543, 'checkpoint_manager_blocking_duration_secs': 1.0463814735412598} I0425 13:36:43.276721 138079697409856 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0425 13:36:43.276773 138079697409856 max_utils.py:750] Memstats: After params initialized: I0425 13:36:43.276819 138079697409856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0425 13:36:43.276851 138079697409856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0425 13:36:43.276878 138079697409856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0425 13:36:43.276902 138079697409856 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0425 13:36:43.585200 138079697409856 metric_logger.py:196] completed step: 10, seconds: 11.370, TFLOP/s/device: 1.195, Tokens/s/device: 180.125, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0425 13:36:44.191880 138079697409856 metric_logger.py:196] completed step: 11, seconds: 1.553, TFLOP/s/device: 8.750, Tokens/s/device: 1318.968, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0425 13:36:44.339241 138079697409856 metric_logger.py:196] completed step: 12, seconds: 0.460, TFLOP/s/device: 29.561, Tokens/s/device: 4455.787, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0425 13:36:44.486486 138079697409856 metric_logger.py:196] completed step: 13, seconds: 0.153, TFLOP/s/device: 88.740, Tokens/s/device: 13375.829, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0425 13:36:44.491172 138079697409856 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0425 13:36:44.780729 138079697409856 checkpointing.py:798] Waited 0.28952693939208984 seconds for step 15 to finish before starting checkpointing. I0425 13:36:44.783825 138079697409856 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 13:36:48.800571 137959223314176 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6 I0425 13:37:25.022875 137961370797824 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.577 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.036524295806885 s) (per-host) I0425 13:37:25.022996 137961370797824 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.752210s. I0425 13:37:33.405838 137961370797824 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 50.135037s. I0425 13:37:33.406125 137959214921472 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 13:37:33.406245 137959214921472 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 13:37:33.406293 137959214921472 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0425 13:37:33.408414 137959214921472 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0425 13:37:33.408601 138079697409856 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=10][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=10. W0425 13:37:33.408745 138079697409856 checkpoint_manager.py:1452] Waiting for previous save to complete took 48.624928 seconds. If this number is high, consider checkpointing less frequently. I0425 13:37:33.411001 138079697409856 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0425 13:37:33.413527 138079697409856 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15. I0425 13:37:33.716529 138079697409856 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 13:37:33.716621 138079697409856 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 I0425 13:37:33.760341 138079697409856 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.044981s I0425 13:37:33.760507 138079697409856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 31.916 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04833173751831055 s) (per-host) I0425 13:37:33.760560 138079697409856 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.048393s (batch_requests_ready=0.001725s, total_serialization_initiated=0.046593s, others=0.000075s) I0425 13:37:33.760646 138079697409856 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.053358s (all_items=0.000013s, per_item={'items': '0.00001311'}, temp_paths=0.053345) I0425 13:37:33.761349 138079697409856 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15. I0425 13:37:33.761672 137959223314176 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 13:57:33.761639 I0425 13:37:33.763609 138079697409856 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0425 13:37:33.763843 137959214921472 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 13:37:33.763967 138079697409856 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777124204.7837887, 'wait_for_prev_duration_secs': 48.62492752075195, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777124253.4110413, 'checkpointer_blocking_duration_secs': 0.3507719039916992, 'get_old_steps_start_time': 1777124253.7618365, 'get_old_steps_duration_secs': 3.314018249511719e-05, 'checkpoint_manager_blocking_start_time': 1777124204.7813334, 'checkpoint_manager_blocking_duration_secs': 48.98260188102722} I0425 13:37:33.764072 138079697409856 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0425 13:37:33.764895 138079697409856 metric_logger.py:196] completed step: 14, seconds: 0.147, TFLOP/s/device: 92.227, Tokens/s/device: 13901.522, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0425 13:37:33.770622 138079697409856 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.224, Tokens/s/device: 13900.956, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0425 13:37:38.844374 137955456825088 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6 I0425 13:37:54.678704 138079697409856 metric_logger.py:196] completed step: 16, seconds: 49.278, TFLOP/s/device: 0.276, Tokens/s/device: 41.560, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0425 13:37:54.826155 138079697409856 metric_logger.py:196] completed step: 17, seconds: 20.762, TFLOP/s/device: 0.654, Tokens/s/device: 98.643, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0425 13:37:54.973489 138079697409856 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.117, Tokens/s/device: 13432.679, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0425 13:37:55.120144 138079697409856 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0425 13:37:55.120830 138079697409856 checkpointing.py:798] Waited 0.0007045269012451172 seconds for step 19 to finish before starting checkpointing. I0425 13:37:55.123080 138079697409856 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 13:38:15.236155 137959223314176 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.041 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.52393293380737 s) (per-host) I0425 13:38:15.236260 137959223314176 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.474474s. I0425 13:38:22.183368 137959223314176 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.421566s. I0425 13:38:22.183681 137959214921472 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 13:38:22.183826 137959214921472 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 13:38:22.183884 137959214921472 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0425 13:38:22.185642 137959214921472 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0425 13:38:22.185830 138079697409856 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=15][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=15. W0425 13:38:22.185950 138079697409856 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.062877 seconds. If this number is high, consider checkpointing less frequently. I0425 13:38:22.187587 138079697409856 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0425 13:38:22.189697 138079697409856 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19. I0425 13:38:22.473990 138079697409856 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0425 13:38:22.474079 138079697409856 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 I0425 13:38:22.515090 138079697409856 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.041925s I0425 13:38:22.515263 138079697409856 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 34.026 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04533505439758301 s) (per-host) I0425 13:38:22.515312 138079697409856 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045393s (batch_requests_ready=0.001638s, total_serialization_initiated=0.043692s, others=0.000062s) I0425 13:38:22.515404 138079697409856 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.049457s (all_items=0.000012s, per_item={'items': '0.00001216'}, temp_paths=0.049444) I0425 13:38:22.516081 138079697409856 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.33 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19. I0425 13:38:22.516437 137955456825088 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-25 13:58:22.516401 I0425 13:38:22.518330 138079697409856 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0425 13:38:22.518597 137959214921472 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0425 13:38:22.518748 138079697409856 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777124275.1230512, 'wait_for_prev_duration_secs': 27.062877416610718, 'time_between_consecutive_saves_sec': 21.71459412574768, 'checkpointer_blocking_start_time': 1777124302.1876287, 'checkpointer_blocking_duration_secs': 0.32891178131103516, 'get_old_steps_start_time': 1777124302.5165594, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1777124275.1210537, 'checkpoint_manager_blocking_duration_secs': 27.397664308547974} I0425 13:38:22.518855 138079697409856 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0425 13:38:22.518898 138079697409856 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0425 13:38:28.073130 137955448432384 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260425_121355/linen_xpk_feat_nnx_post_train_fixes_20260425_121355_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6 I0425 13:39:04.089534 137955456825088 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.953 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.61957025527954 s) (per-host) I0425 13:39:04.089641 137955456825088 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.572592s. I0425 13:39:11.548434 137955456825088 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.031370s. I0425 13:39:11.548732 137959214921472 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0425 13:39:11.548855 137959214921472 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0425 13:39:11.548895 137959214921472 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0425 13:39:11.550460 137959214921472 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0425 13:39:11.550645 138079697409856 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=19][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=19. I0425 13:39:11.550792 138079697409856 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0425 13:39:11.551859 138079697409856 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.224, Tokens/s/device: 13901.050, 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: Sat Apr 25 13:39:22 UTC 2026 EXIT_CODE=0