XPK Start: Fri Apr 24 13:28:39 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. 2026-04-24 13:29:04.370366: 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) I0424 13:29:04.583622 139196182087488 max_utils.py:273] Attempting to initialize the jax distributed system... I0424 13:29:13.625763 139196182087488 distributed.py:149] Starting JAX distributed service on [::]:8482 I0424 13:29:13.628122 139196182087488 distributed.py:172] Connecting to JAX distributed service on mt-16-zarr3-false-rs-ts5iw-slice-job-0-0.mt-16-zarr3-false-rs-ts5iw:8482 I0424 13:29:14.542855 139196182087488 max_utils.py:284] Jax distributed system initialized! I0424 13:29:20.901869 139196182087488 max_utils.py:800] System Information: Jax Version: 0.9.2 I0424 13:29:20.901978 139196182087488 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0424 13:29:20.902019 139196182087488 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 I0424 13:29:20.902055 139196182087488 train_utils.py:391] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0424 13:29:21.598716 139196182087488 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 13:29:21.599307 139196182087488 maxtext_utils.py:1771] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0424 13:29:21.599488 139196182087488 checkpointing.py:688] Setting up checkpoint logger... I0424 13:29:21.599543 139196182087488 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=False I0424 13:29:21.599585 139196182087488 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0424 13:29:21.599920 139196182087488 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 0x7e989b7183e0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0424 13:29:22.944565 139196182087488 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0424 13:29:22.944791 139196182087488 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 0x7e8478176f90>}, handler_registry=None I0424 13:29:22.945029 139196182087488 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e8478176f90>` 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`. I0424 13:29:22.945077 139196182087488 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e84607233e0>` 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`. I0424 13:29:22.945128 139196182087488 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 0x7e8478176f90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e8478176f90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e84607233e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e84607233e0>}). I0424 13:29:22.945456 139196182087488 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0424 13:29:22.945533 139196182087488 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e84602f1d00> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0424 13:29:24.478590 139075319224064 checkpoint.py:228] Read Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1777037095664177392, 'commit_timestamp_nsecs': 1777037142115111419, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/0/_CHECKPOINT_METADATA I0424 13:29:24.734747 139075319224064 checkpoint_manager.py:1871] Missing metrics for step 0 I0424 13:29:24.929973 139196182087488 checkpoint_manager.py:1812] Found 3 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints I0424 13:29:25.113546 139196182087488 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_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e84607212b0> I0424 13:29:25.113720 139196182087488 checkpointing.py:302] Checkpoint manager created! I0424 13:29:26.055480 139196182087488 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0424 13:29:26.055626 139196182087488 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 13:29:26.436248 139196182087488 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 13:29:26.436339 139196182087488 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 13:29:26.452766 139196182087488 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch_attn', 'activation_length_attn', 'activation_embed_attn'). I0424 13:29:26.452825 139196182087488 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0424 13:29:26.476740 139196182087488 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 13:29:26.476809 139196182087488 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 13:29:26.493397 139196182087488 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 13:29:26.493460 139196182087488 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 13:29:26.509992 139196182087488 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_length_attn', 'activation_kv_heads', 'activation_kv_head_dim'). I0424 13:29:26.510053 139196182087488 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 13:29:26.535046 139196182087488 attentions.py:1198] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch_attn', 'activation_length_attn', 'activation_heads', 'activation_kv'). I0424 13:29:26.535127 139196182087488 attentions.py:1198] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0424 13:29:26.556105 139196182087488 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0424 13:29:26.556169 139196182087488 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0424 13:29:26.780026 139196182087488 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0424 13:29:26.780154 139196182087488 checkpointing.py:582] restoring from this run's directory step 9 I0424 13:29:27.026653 139196182087488 event_tracking.py:70] [process=6] [sync] Started load checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 13:29:27.273001 139196182087488 checkpointer.py:307] Restoring checkpoint from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 13:29:27.273460 139196182087488 event_tracking.py:125] [process=6] [sync] Finished blocking load in 0.25 seconds. Continuing load @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9. I0424 13:29:28.692066 1449 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0424 13:29:34.890985 139196182087488 jax_array_handlers.py:843] [process=6] /jax/orbax/read/worker/io/requested throughput: 254.687 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.204761028289795 s) (per-host) I0424 13:29:34.895681 139196182087488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/checkpoint/read/gbytes_per_sec: 239.182 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 6.616701126098633 s) (per-host) I0424 13:29:35.875079 139196182087488 event_tracking.py:138] [process=6] [sync] Finished load in 8.85 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/9 I0424 13:29:35.875294 139196182087488 standard_logger.py:34] {'step': 9, 'event_type': 'restore', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'checkpointer_start_time': 1777037367.0265572, 'checkpointer_duration_secs': 8.848704814910889, 'checkpoint_manager_start_time': 1777037366.780413, 'checkpoint_manager_duration_secs': 9.094850778579712} fsdp: 32 I0424 13:29:36.290319 139196182087488 maxtext_utils.py:1880] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0424 13:29:36.290452 139196182087488 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) I0424 13:29:36.290509 139196182087488 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) I0424 13:29:36.290568 139196182087488 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') I0424 13:29:36.290629 139196182087488 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) I0424 13:29:36.290670 139196182087488 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) I0424 13:29:36.290723 139196182087488 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) I0424 13:29:36.290777 139196182087488 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') I0424 13:29:36.290817 139196182087488 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) I0424 13:29:36.290855 139196182087488 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) I0424 13:29:36.290903 139196182087488 maxtext_utils.py:1880] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0424 13:29:36.290951 139196182087488 maxtext_utils.py:1880] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0424 13:29:36.777178 139196182087488 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 13:29:36.777269 139196182087488 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0424 13:29:36.792825 139196182087488 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0424 13:29:36.792886 139196182087488 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0424 13:29:47.877327 139196182087488 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. I0424 13:29:47.878157 139196182087488 metric_logger.py:301] number parameters: 1.104 billion I0424 13:29:59.160339 139196182087488 checkpointing.py:794] Waiting for step 10 to finish before checkpoint... I0424 13:29:59.314517 139196182087488 checkpointing.py:798] Waited 0.1541576385498047 seconds for step 10 to finish before starting checkpointing. I0424 13:29:59.316848 139196182087488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 13:29:59.318861 139196182087488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 10 I0424 13:29:59.320339 139196182087488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10. I0424 13:29:59.660171 139196182087488 signaling_client.py:364] Using JaxDistributedSignalingClient I0424 13:29:59.661056 139196182087488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0424 13:29:59.661129 139196182087488 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 I0424 13:29:59.933679 139196182087488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.273596s I0424 13:29:59.933856 139196182087488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 5.559 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.277482271194458 s) (per-host) I0424 13:29:59.933911 139196182087488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.277548s (batch_requests_ready=0.002184s, total_serialization_initiated=0.275290s, others=0.000074s) I0424 13:29:59.934008 139196182087488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.281765s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.281747) I0424 13:29:59.934831 139196182087488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.62 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10. I0424 13:29:59.935157 139074715244288 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 13:49:59.935119 I0424 13:29:59.947585 139196182087488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=10] Starting CheckpointManager Save Finalize thread=save_finalize I0424 13:29:59.947841 139073121416960 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 13:29:59.947992 139196182087488 standard_logger.py:34] {'step': 10, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777037399.3168292, 'wait_for_prev_duration_secs': 6.031990051269531e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777037399.3188992, 'checkpointer_blocking_duration_secs': 0.61641526222229, 'get_old_steps_start_time': 1777037399.9353383, 'get_old_steps_duration_secs': 3.409385681152344e-05, 'checkpoint_manager_blocking_start_time': 1777037399.314899, 'checkpoint_manager_blocking_duration_secs': 0.6330597400665283} I0424 13:29:59.948108 139196182087488 checkpointing.py:409] Started an asynchronous checkpoint save for step 10 I0424 13:29:59.948155 139196182087488 max_utils.py:750] Memstats: After params initialized: I0424 13:29:59.948205 139196182087488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0424 13:29:59.948238 139196182087488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0424 13:29:59.948265 139196182087488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0424 13:29:59.948288 139196182087488 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0424 13:30:00.258046 139196182087488 metric_logger.py:196] completed step: 10, seconds: 11.282, TFLOP/s/device: 1.204, Tokens/s/device: 181.527, total_weights: 65536, loss: 6.833, lm_loss: 6.833, perplexity: 927.869 I0424 13:30:00.850965 139196182087488 metric_logger.py:196] completed step: 11, seconds: 1.096, TFLOP/s/device: 12.395, Tokens/s/device: 1868.236, total_weights: 65536, loss: 6.527, lm_loss: 6.527, perplexity: 683.659 I0424 13:30:00.998736 139196182087488 metric_logger.py:196] completed step: 12, seconds: 0.447, TFLOP/s/device: 30.402, Tokens/s/device: 4582.517, total_weights: 65536, loss: 6.265, lm_loss: 6.265, perplexity: 526.068 I0424 13:30:01.146180 139196182087488 metric_logger.py:196] completed step: 13, seconds: 0.152, TFLOP/s/device: 89.448, Tokens/s/device: 13482.643, total_weights: 65536, loss: 6.045, lm_loss: 6.045, perplexity: 422.001 I0424 13:30:01.150592 139196182087488 checkpointing.py:794] Waiting for step 15 to finish before checkpoint... I0424 13:30:01.440258 139196182087488 checkpointing.py:798] Waited 0.2896265983581543 seconds for step 15 to finish before starting checkpointing. I0424 13:30:01.445445 139196182087488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=10][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 13:30:05.607642 139067178080000 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_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/10/items/array_metadatas/process_6 I0424 13:30:40.992997 139074715244288 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.213 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.33659291267395 s) (per-host) I0424 13:30:40.993122 139074715244288 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.057853s. I0424 13:30:48.919284 139074715244288 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 48.983997s. I0424 13:30:48.919572 139073121416960 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 13:30:48.919699 139073121416960 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 13:30:48.919748 139073121416960 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is syncing with other hosts... I0424 13:30:48.921206 139073121416960 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=10] CheckpointManager Save Finalize is done on all hosts. I0424 13:30:48.921412 139196182087488 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. W0424 13:30:48.921542 139196182087488 checkpoint_manager.py:1452] Waiting for previous save to complete took 47.476097 seconds. If this number is high, consider checkpointing less frequently. I0424 13:30:48.923811 139196182087488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 15 I0424 13:30:48.926357 139196182087488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15. I0424 13:30:49.237993 139196182087488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0424 13:30:49.238086 139196182087488 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 I0424 13:30:49.279420 139196182087488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042500s I0424 13:30:49.279590 139196182087488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 33.586 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.0459287166595459 s) (per-host) I0424 13:30:49.279645 139196182087488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.045994s (batch_requests_ready=0.001775s, total_serialization_initiated=0.044148s, others=0.000071s) I0424 13:30:49.279731 139196182087488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.051075s (all_items=0.000014s, per_item={'items': '0.00001383'}, temp_paths=0.051062) I0424 13:30:49.280377 139196182087488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.36 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15. I0424 13:30:49.280667 139067178080000 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 13:50:49.280631 I0424 13:30:49.282495 139196182087488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=15] Starting CheckpointManager Save Finalize thread=save_finalize I0424 13:30:49.282784 139073121416960 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 13:30:49.282942 139196182087488 standard_logger.py:34] {'step': 15, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777037401.4454165, 'wait_for_prev_duration_secs': 47.476096868515015, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1777037448.9238513, 'checkpointer_blocking_duration_secs': 0.35695576667785645, 'get_old_steps_start_time': 1777037449.2808273, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1777037401.4408627, 'checkpoint_manager_blocking_duration_secs': 47.842047452926636} I0424 13:30:49.283047 139196182087488 checkpointing.py:409] Started an asynchronous checkpoint save for step 15 I0424 13:30:49.283869 139196182087488 metric_logger.py:196] completed step: 14, seconds: 0.148, TFLOP/s/device: 92.009, Tokens/s/device: 13868.668, total_weights: 65536, loss: 5.864, lm_loss: 5.864, perplexity: 352.193 I0424 13:30:49.289778 139196182087488 metric_logger.py:196] completed step: 15, seconds: 0.147, TFLOP/s/device: 92.155, Tokens/s/device: 13890.585, total_weights: 65536, loss: 5.719, lm_loss: 5.719, perplexity: 304.661 I0424 13:30:54.652214 139070959429376 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_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/15/items/array_metadatas/process_6 I0424 13:31:10.562665 139196182087488 metric_logger.py:196] completed step: 16, seconds: 48.138, TFLOP/s/device: 0.282, Tokens/s/device: 42.545, total_weights: 65536, loss: 5.605, lm_loss: 5.605, perplexity: 271.872 I0424 13:31:10.710181 139196182087488 metric_logger.py:196] completed step: 17, seconds: 21.127, TFLOP/s/device: 0.643, Tokens/s/device: 96.939, total_weights: 65536, loss: 5.517, lm_loss: 5.517, perplexity: 248.975 I0424 13:31:10.857392 139196182087488 metric_logger.py:196] completed step: 18, seconds: 0.152, TFLOP/s/device: 89.214, Tokens/s/device: 13447.232, total_weights: 65536, loss: 5.448, lm_loss: 5.448, perplexity: 232.333 I0424 13:31:11.003984 139196182087488 checkpointing.py:794] Waiting for step 19 to finish before checkpoint... I0424 13:31:11.004661 139196182087488 checkpointing.py:798] Waited 0.0006985664367675781 seconds for step 19 to finish before starting checkpointing. I0424 13:31:11.006708 139196182087488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=15][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 13:31:30.886848 139067178080000 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 37.923 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.65314483642578 s) (per-host) I0424 13:31:30.886973 139067178080000 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.606194s. I0424 13:31:38.557671 139067178080000 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.276876s. I0424 13:31:38.557939 139073121416960 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 13:31:38.558062 139073121416960 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 13:31:38.558137 139073121416960 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is syncing with other hosts... I0424 13:31:38.559705 139073121416960 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=15] CheckpointManager Save Finalize is done on all hosts. I0424 13:31:38.559876 139196182087488 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. W0424 13:31:38.559996 139196182087488 checkpoint_manager.py:1452] Waiting for previous save to complete took 27.553294 seconds. If this number is high, consider checkpointing less frequently. I0424 13:31:38.561541 139196182087488 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 19 I0424 13:31:38.563667 139196182087488 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19. I0424 13:31:38.886535 139196182087488 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0424 13:31:38.886624 139196182087488 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 I0424 13:31:38.928411 139196182087488 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.042869s I0424 13:31:38.928570 139196182087488 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 33.345 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.04626107215881348 s) (per-host) I0424 13:31:38.928619 139196182087488 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.046320s (batch_requests_ready=0.001629s, total_serialization_initiated=0.044627s, others=0.000063s) I0424 13:31:38.928702 139196182087488 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.050359s (all_items=0.000012s, per_item={'items': '0.00001168'}, temp_paths=0.050347) I0424 13:31:38.929450 139196182087488 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.37 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19. I0424 13:31:38.929734 139073121416960 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-24 13:51:38.929710 I0424 13:31:38.931617 139196182087488 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=19] Starting CheckpointManager Save Finalize thread=save_finalize I0424 13:31:38.931828 139070959429376 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0424 13:31:38.931977 139196182087488 standard_logger.py:34] {'step': 19, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1777037471.0066795, 'wait_for_prev_duration_secs': 27.55329418182373, 'time_between_consecutive_saves_sec': 22.085434198379517, 'checkpointer_blocking_start_time': 1777037498.5615797, 'checkpointer_blocking_duration_secs': 0.3682565689086914, 'get_old_steps_start_time': 1777037498.9298556, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1777037471.0048919, 'checkpoint_manager_blocking_duration_secs': 27.927055597305298} I0424 13:31:38.932084 139196182087488 checkpointing.py:409] Started an asynchronous checkpoint save for step 19 I0424 13:31:38.932141 139196182087488 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=19][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0424 13:31:44.337431 139070951036672 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_20260424_120657/linen_xpk_feat_nnx_post_train_fixes_20260424_120657_16_zarr3_false/checkpoints/19/items/array_metadatas/process_6 I0424 13:32:19.707242 139073121416960 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 38.692 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.82489061355591 s) (per-host) I0424 13:32:19.707363 139073121416960 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.777549s. I0424 13:32:28.171308 139073121416960 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 49.241477s. I0424 13:32:28.171610 139070959429376 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0424 13:32:28.171730 139070959429376 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0424 13:32:28.171773 139070959429376 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is syncing with other hosts... I0424 13:32:28.174237 139070959429376 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=19] CheckpointManager Save Finalize is done on all hosts. I0424 13:32:28.174396 139196182087488 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. I0424 13:32:28.174540 139196182087488 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0424 13:32:28.175437 139196182087488 metric_logger.py:196] completed step: 19, seconds: 0.147, TFLOP/s/device: 92.199, Tokens/s/device: 13897.183, 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: Fri Apr 24 13:32:40 UTC 2026 EXIT_CODE=0