XPK Start: Mon Apr 20 12:25:06 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 12:25:31.013568: 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 12:25:31.192499 138158368888640 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 12:25:40.233339 138158368888640 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 12:25:40.235659 138158368888640 distributed.py:172] Connecting to JAX distributed service on mt-04-int8-kw1ya-slice-job-0-0.mt-04-int8-kw1ya:8482 I0420 12:25:41.836788 138158368888640 max_utils.py:284] Jax distributed system initialized! I0420 12:25:47.627608 138158368888640 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 12:25:47.627711 138158368888640 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 12:25:47.627750 138158368888640 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 12:25:47.627784 138158368888640 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 12:25:48.664127 138158368888640 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 12:25:48.664711 138158368888640 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 12:25:48.665002 138158368888640 checkpointing.py:688] Setting up checkpoint logger... I0420 12:25:48.665056 138158368888640 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 12:25:48.665110 138158368888640 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 12:25:48.665427 138158368888640 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7da6e2914560>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 12:25:51.555983 138158368888640 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 12:25:51.556334 138158368888640 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7da5bcf71580>}, handler_registry=None I0420 12:25:51.556574 138158368888640 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7da5bcf71580>` 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 12:25:51.556630 138158368888640 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d932c453cb0>` 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 12:25:51.556665 138158368888640 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 0x7da5bcf71580>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7da5bcf71580>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d932c453cb0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7d932c453cb0>}). I0420 12:25:51.556974 138158368888640 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 12:25:51.557042 138158368888640 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7d932c23ed40> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 12:25:52.911998 138158368888640 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints I0420 12:25:52.914314 138158368888640 checkpoint_manager.py:929] [process=0][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=10), 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_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7da5bcf80c50> I0420 12:25:52.914425 138158368888640 checkpointing.py:302] Checkpoint manager created! I0420 12:25:54.036611 138158368888640 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 12:25:54.036712 138158368888640 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 12:25:54.419004 138158368888640 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 12:25:54.419105 138158368888640 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 12:25:54.435587 138158368888640 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 12:25:54.435647 138158368888640 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0420 12:25:54.504995 138158368888640 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 12:25:54.505102 138158368888640 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 12:25:54.521419 138158368888640 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 12:25:54.521480 138158368888640 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 12:25:54.537618 138158368888640 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 12:25:54.537693 138158368888640 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 12:25:54.562180 138158368888640 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 12:25:54.562248 138158368888640 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0420 12:25:54.627477 138158368888640 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 12:25:54.627565 138158368888640 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0420 12:25:55.139695 138158368888640 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 12:25:55.139803 138158368888640 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0420 12:25:57.177338 138158368888640 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0420 12:25:57.177459 138158368888640 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 12:25:57.177521 138158368888640 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 12:25:57.177583 138158368888640 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 12:25:57.177641 138158368888640 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 12:25:57.177681 138158368888640 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 12:25:57.177733 138158368888640 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 12:25:57.177786 138158368888640 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 12:25:57.177826 138158368888640 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 12:25:57.177863 138158368888640 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 12:25:57.177911 138158368888640 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0420 12:25:57.177958 138158368888640 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0420 12:25:59.012765 138158368888640 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 12:25:59.012858 138158368888640 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0420 12:25:59.028380 138158368888640 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0420 12:25:59.028441 138158368888640 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0420 12:26:12.309371 138158368888640 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.4 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0420 12:26:12.317003 138158368888640 metric_logger.py:301] number parameters: 1.104 billion I0420 12:26:27.509212 138158368888640 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 12:26:27.749661 138158368888640 checkpointing.py:798] Waited 0.24042654037475586 seconds for step 0 to finish before starting checkpointing. I0420 12:26:27.752330 138158368888640 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 12:26:27.754526 138158368888640 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 12:26:27.755939 138158368888640 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0. I0420 12:26:28.458675 138158368888640 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 12:26:28.474021 138158368888640 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 12:26:28.474129 138158368888640 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 12:26:28.722452 138029796173568 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0 I0420 12:26:28.757594 138158368888640 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.284403s I0420 12:26:28.758526 138158368888640 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.304 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2908504009246826 s) (per-host) I0420 12:26:28.758615 138158368888640 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.290974s (batch_requests_ready=0.002101s, total_serialization_initiated=0.288012s, others=0.000860s) I0420 12:26:28.758721 138158368888640 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.295708s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.295691) I0420 12:26:28.759618 138158368888640 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.00 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0. I0420 12:26:28.760015 138029811300096 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 12:46:28.759975 I0420 12:26:28.765639 138158368888640 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 12:26:28.765933 138028727125760 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 12:26:28.766099 138158368888640 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776687987.7523127, 'wait_for_prev_duration_secs': 5.91278076171875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776687987.7545657, 'checkpointer_blocking_duration_secs': 1.0056214332580566, 'get_old_steps_start_time': 1776687988.760212, 'get_old_steps_duration_secs': 2.8371810913085938e-05, 'checkpoint_manager_blocking_start_time': 1776687987.7502272, 'checkpoint_manager_blocking_duration_secs': 1.015824317932129} I0420 12:26:28.766200 138158368888640 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 12:26:28.766251 138158368888640 max_utils.py:750] Memstats: After params initialized: I0420 12:26:28.766297 138158368888640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 12:26:28.766329 138158368888640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 12:26:28.766357 138158368888640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 12:26:28.766380 138158368888640 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 12:26:29.113260 138158368888640 metric_logger.py:196] completed step: 0, seconds: 14.881, TFLOP/s/device: 0.913, Tokens/s/device: 137.621, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277 I0420 12:26:29.116436 138158368888640 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/tensorboard/' I0420 12:26:29.543166 138158368888640 metric_logger.py:196] completed step: 1, seconds: 1.602, TFLOP/s/device: 8.481, Tokens/s/device: 1278.344, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52796.277 I0420 12:26:29.770674 138158368888640 metric_logger.py:196] completed step: 2, seconds: 0.431, TFLOP/s/device: 31.550, Tokens/s/device: 4755.547, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28634.820 I0420 12:26:29.999988 138158368888640 metric_logger.py:196] completed step: 3, seconds: 0.010, TFLOP/s/device: 1398.140, Tokens/s/device: 210742.951, total_weights: 65536, loss: 9.731, lm_loss: 9.731, perplexity: 16827.199 I0420 12:26:30.521139 138029301753600 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/items I0420 12:26:30.665441 138027650266880 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776687990260592366, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776687990260592366, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0420 12:26:32.084015 2870 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 12:26:33.875953 138028743911168 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/items/array_metadatas/process_0 I0420 12:27:00.005524 138158368888640 metric_logger.py:196] completed step: 4, seconds: 0.228, TFLOP/s/device: 59.522, Tokens/s/device: 8971.871, total_weights: 65536, loss: 9.272, lm_loss: 9.272, perplexity: 10638.531 I0420 12:27:00.014236 138158368888640 metric_logger.py:196] completed step: 5, seconds: 0.230, TFLOP/s/device: 59.005, Tokens/s/device: 8893.907, total_weights: 65536, loss: 8.887, lm_loss: 8.887, perplexity: 7233.941 I0420 12:27:00.241620 138158368888640 metric_logger.py:196] completed step: 6, seconds: 30.003, TFLOP/s/device: 0.453, Tokens/s/device: 68.260, total_weights: 65536, loss: 8.588, lm_loss: 8.588, perplexity: 5365.468 I0420 12:27:00.472641 138158368888640 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 2031.265, Tokens/s/device: 306174.316, total_weights: 65536, loss: 8.380, lm_loss: 8.380, perplexity: 4358.779 I0420 12:27:00.703286 138158368888640 metric_logger.py:196] completed step: 8, seconds: 0.228, TFLOP/s/device: 59.493, Tokens/s/device: 8967.471, total_weights: 65536, loss: 8.251, lm_loss: 8.251, perplexity: 3831.453 I0420 12:27:00.933209 138158368888640 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 12:27:00.933759 138158368888640 checkpointing.py:798] Waited 0.0005719661712646484 seconds for step 9 to finish before starting checkpointing. I0420 12:27:00.936004 138158368888640 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 12:27:05.219325 138028735518464 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.140008s (commit=31.941600s, array_metadata_write=1.198407s) I0420 12:27:05.221857 138029811300096 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 42.977 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.75416803359985 s) (per-host) I0420 12:27:05.221976 138029811300096 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.461823s. I0420 12:27:05.979386 138029811300096 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776687990260592366, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0420 12:27:06.725005 138027650266880 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776687990260592366, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/_CHECKPOINT_METADATA I0420 12:27:07.192872 138029811300096 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000266s. I0420 12:27:07.928973 138029811300096 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 12:27:08.865453 138029811300096 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.766083s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/items I0420 12:27:08.867334 138029811300096 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0/items I0420 12:27:09.377883 138029811300096 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0 I0420 12:27:10.817239 138029811300096 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0`. I0420 12:27:10.817991 138029811300096 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 43.06 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/0 I0420 12:27:10.819575 138029811300096 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 42.059407s. I0420 12:27:10.819731 138028727125760 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 12:27:10.819792 138028727125760 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 12:27:10.819861 138028727125760 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 12:27:10.822981 138028727125760 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 12:27:10.823115 138158368888640 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 12:27:10.823203 138158368888640 checkpoint_manager.py:1452] Waiting for previous save to complete took 9.887203 seconds. If this number is high, consider checkpointing less frequently. I0420 12:27:10.825209 138158368888640 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0420 12:27:10.827274 138158368888640 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9. I0420 12:27:11.143039 138158368888640 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 12:27:11.143140 138158368888640 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 12:27:11.169414 138158368888640 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.027109s I0420 12:27:11.170330 138158368888640 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 48.518 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.031794071197509766 s) (per-host) I0420 12:27:11.170391 138158368888640 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.031870s (batch_requests_ready=0.001837s, total_serialization_initiated=0.029210s, others=0.000823s) I0420 12:27:11.170463 138158368888640 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.035915s (all_items=0.000015s, per_item={'items': '0.00001502'}, temp_paths=0.035900) I0420 12:27:11.171221 138158368888640 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.35 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9. I0420 12:27:11.171553 138028727125760 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 12:47:11.171517 I0420 12:27:11.179460 138158368888640 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 12:27:11.179774 138027154269952 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 12:27:11.179945 138158368888640 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776688020.935973, 'wait_for_prev_duration_secs': 9.887203216552734, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776688030.8252473, 'checkpointer_blocking_duration_secs': 0.34644627571105957, 'get_old_steps_start_time': 1776688031.171719, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776688020.9339921, 'checkpoint_manager_blocking_duration_secs': 10.245919942855835} I0420 12:27:11.180064 138158368888640 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 12:27:11.180122 138158368888640 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 12:27:11.376753 138029811300096 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9 I0420 12:27:13.542363 138029268182784 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9/items I0420 12:27:16.739133 138028195936000 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9/items/array_metadatas/process_0 I0420 12:27:56.744405 138027162662656 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.073659s (commit=40.967458s, array_metadata_write=1.106202s) I0420 12:27:56.745436 138028727125760 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.635 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.60685920715332 s) (per-host) I0420 12:27:56.745572 138028727125760 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.573904s. I0420 12:27:58.177792 138028727125760 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000224s. I0420 12:27:58.831043 138028727125760 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 12:27:59.771520 138028727125760 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.178846s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9/items I0420 12:27:59.773215 138028727125760 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9/items I0420 12:28:00.296941 138028727125760 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9 I0420 12:28:01.712149 138028727125760 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9`. I0420 12:28:01.712847 138028727125760 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.89 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158/linen_xpk_feat_nnx_linen_converter_and_sharding_tools_20260420_112158_04_int8/checkpoints/9 I0420 12:28:01.714393 138028727125760 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.542724s. I0420 12:28:01.714606 138027154269952 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 12:28:01.714728 138027154269952 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 12:28:01.714789 138027154269952 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 12:28:01.716557 138027154269952 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 12:28:01.716723 138158368888640 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 12:28:01.716865 138158368888640 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 12:28:01.717866 138158368888640 metric_logger.py:196] completed step: 9, seconds: 0.231, TFLOP/s/device: 58.891, Tokens/s/device: 8876.676, total_weights: 65536, loss: 8.175, lm_loss: 8.175, perplexity: 3552.333 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 12:28:11 UTC 2026 EXIT_CODE=0