XPK Start: Mon Apr 20 16:17:41 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 16:18:05.796705: 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 16:18:05.974159 132767335823168 max_utils.py:273] Attempting to initialize the jax distributed system... I0420 16:18:15.014887 132767335823168 distributed.py:149] Starting JAX distributed service on [::]:8482 I0420 16:18:15.017216 132767335823168 distributed.py:172] Connecting to JAX distributed service on mt-09-pdb-lt-1-mn02l-slice-job-0-0.mt-09-pdb-lt-1-mn02l:8482 I0420 16:18:16.694577 132767335823168 max_utils.py:284] Jax distributed system initialized! I0420 16:18:21.748646 132767335823168 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 16:18:21.748749 132767335823168 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 16:18:21.748790 132767335823168 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 16:18:21.748826 132767335823168 train_utils.py:378] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0420 16:18:22.436942 132767335823168 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0420 16:18:22.437531 132767335823168 maxtext_utils.py:1718] Num_devices: 32, shape (1, 1, 1, 8, 1, 1, 1, 1, 4, 1, 1, 1, 1) I0420 16:18:22.437712 132767335823168 checkpointing.py:688] Setting up checkpoint logger... I0420 16:18:22.437761 132767335823168 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 16:18:22.437811 132767335823168 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0420 16:18:22.438127 132767335823168 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 0x78bfb9914c20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 16:18:25.411613 132767335823168 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 16:18:25.411850 132767335823168 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 0x78ac041a9dc0>}, handler_registry=None I0420 16:18:25.412108 132767335823168 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78ac041a9dc0>` 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 16:18:25.412163 132767335823168 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ac041adb50>` 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 16:18:25.412199 132767335823168 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 0x78ac041a9dc0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x78ac041a9dc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ac041adb50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x78ac041adb50>}). I0420 16:18:25.412506 132767335823168 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0420 16:18:25.412575 132767335823168 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x78ab1c29b740> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0420 16:18:26.633980 132767335823168 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints I0420 16:18:27.025030 132767335823168 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x78ac041a9070> I0420 16:18:27.025234 132767335823168 checkpointing.py:302] Checkpoint manager created! I0420 16:18:27.940900 132767335823168 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 16:18:27.940999 132767335823168 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 16:18:28.324010 132767335823168 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 16:18:28.324110 132767335823168 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 16:18:28.340477 132767335823168 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 16:18:28.340538 132767335823168 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, 'tensor'). I0420 16:18:28.363945 132767335823168 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:18:28.364007 132767335823168 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 16:18:28.380352 132767335823168 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:18:28.380413 132767335823168 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 16:18:28.396627 132767335823168 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0420 16:18:28.396684 132767335823168 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 16:18:28.421205 132767335823168 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 16:18:28.421274 132767335823168 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, 'tensor', None). I0420 16:18:28.441931 132767335823168 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 16:18:28.441992 132767335823168 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, 'tensor'). I0420 16:18:28.654600 132767335823168 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 16:18:28.654702 132767335823168 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 8 tensor: 4 I0420 16:18:30.162718 132767335823168 maxtext_utils.py:1827] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: ('tensor',) I0420 16:18:30.162843 132767335823168 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, 'tensor') I0420 16:18:30.162921 132767335823168 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, 'tensor') I0420 16:18:30.162984 132767335823168 maxtext_utils.py:1827] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: ('tensor', None, 'fsdp') I0420 16:18:30.163037 132767335823168 maxtext_utils.py:1827] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: ('tensor', None) I0420 16:18:30.163090 132767335823168 maxtext_utils.py:1827] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: ('tensor', None) I0420 16:18:30.163149 132767335823168 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, 'tensor', None) I0420 16:18:30.163202 132767335823168 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: ('tensor', None, None, 'fsdp') I0420 16:18:30.163243 132767335823168 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, 'tensor', None) I0420 16:18:30.163280 132767335823168 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, 'tensor', None) I0420 16:18:30.163327 132767335823168 maxtext_utils.py:1827] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', 'tensor') I0420 16:18:30.163377 132767335823168 maxtext_utils.py:1827] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: ('tensor', 'fsdp') I0420 16:18:30.654809 132767335823168 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0420 16:18:30.654901 132767335823168 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0420 16:18:30.670230 132767335823168 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0420 16:18:30.670290 132767335823168 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0420 16:18:45.667363 132767335823168 max_utils.py:791] Total memory size: 0.9 GB, Output size: 0.4 GB, Temp size: 0.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0420 16:18:45.675407 132767335823168 metric_logger.py:301] number parameters: 1.104 billion I0420 16:18:47.213632 132767335823168 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 16:19:01.297999 132767335823168 checkpointing.py:798] Waited 14.08434772491455 seconds for step 0 to finish before starting checkpointing. I0420 16:19:01.300300 132767335823168 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 16:19:01.302489 132767335823168 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0 I0420 16:19:01.303878 132767335823168 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0. I0420 16:19:02.053467 132767335823168 signaling_client.py:364] Using JaxDistributedSignalingClient I0420 16:19:02.068692 132767335823168 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 16:19:02.068759 132767335823168 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 16:19:02.329525 132767335823168 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.261601s I0420 16:19:02.330291 132767335823168 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.761 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.26776671409606934 s) (per-host) I0420 16:19:02.330355 132767335823168 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.267847s (batch_requests_ready=0.002055s, total_serialization_initiated=0.265121s, others=0.000671s) I0420 16:19:02.330427 132767335823168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.272244s (all_items=0.000018s, per_item={'items': '0.00001764'}, temp_paths=0.272227) I0420 16:19:02.331292 132767335823168 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.03 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0. I0420 16:19:02.331565 132641694586624 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 16:39:02.331534 I0420 16:19:02.332866 132641686193920 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0 I0420 16:19:02.345297 132767335823168 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 16:19:02.345589 132640633829120 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 16:19:02.345740 132767335823168 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776701941.3002815, 'wait_for_prev_duration_secs': 6.103515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776701941.3025286, 'checkpointer_blocking_duration_secs': 1.0292034149169922, 'get_old_steps_start_time': 1776701942.3317568, 'get_old_steps_duration_secs': 2.6941299438476562e-05, 'checkpoint_manager_blocking_start_time': 1776701941.298548, 'checkpoint_manager_blocking_duration_secs': 1.0471549034118652} I0420 16:19:02.345843 132767335823168 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 16:19:02.345894 132767335823168 max_utils.py:750] Memstats: After params initialized: I0420 16:19:02.345943 132767335823168 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0)) I0420 16:19:02.345977 132767335823168 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0)) I0420 16:19:02.346004 132767335823168 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0)) I0420 16:19:02.346030 132767335823168 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0)) I0420 16:19:02.698898 132767335823168 metric_logger.py:196] completed step: 0, seconds: 1.179, TFLOP/s/device: 2.881, Tokens/s/device: 434.292, total_weights: 16384, loss: 10.862, lm_loss: 10.862, perplexity: 52153.723 I0420 16:19:02.702322 132767335823168 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/tensorboard/' I0420 16:19:03.147808 132767335823168 metric_logger.py:196] completed step: 1, seconds: 15.483, TFLOP/s/device: 0.219, Tokens/s/device: 33.067, total_weights: 16384, loss: 10.862, lm_loss: 10.862, perplexity: 52153.723 I0420 16:19:03.204638 132767335823168 metric_logger.py:196] completed step: 2, seconds: 0.450, TFLOP/s/device: 7.556, Tokens/s/device: 1138.957, total_weights: 16384, loss: 9.763, lm_loss: 9.763, perplexity: 17373.129 I0420 16:19:03.263431 132767335823168 metric_logger.py:196] completed step: 3, seconds: 0.009, TFLOP/s/device: 366.467, Tokens/s/device: 55237.890, total_weights: 16384, loss: 8.823, lm_loss: 8.823, perplexity: 6791.817 I0420 16:19:09.163899 132641198171904 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/items I0420 16:19:09.388171 132640125589248 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701948909136951, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776701948909136951, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0420 16:19:10.764229 2546 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0420 16:19:13.004883 132640650614528 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/items/array_metadatas/process_0 I0420 16:19:18.125545 132767335823168 metric_logger.py:196] completed step: 4, seconds: 0.057, TFLOP/s/device: 59.129, Tokens/s/device: 8912.563, total_weights: 16384, loss: 7.947, lm_loss: 7.947, perplexity: 2828.417 I0420 16:19:18.134580 132767335823168 metric_logger.py:196] completed step: 5, seconds: 0.059, TFLOP/s/device: 57.747, Tokens/s/device: 8704.226, total_weights: 16384, loss: 7.212, lm_loss: 7.212, perplexity: 1355.339 I0420 16:19:18.191555 132767335823168 metric_logger.py:196] completed step: 6, seconds: 14.861, TFLOP/s/device: 0.229, Tokens/s/device: 34.452, total_weights: 16384, loss: 6.652, lm_loss: 6.652, perplexity: 774.659 I0420 16:19:18.250103 132767335823168 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 481.062, Tokens/s/device: 72510.976, total_weights: 16384, loss: 6.269, lm_loss: 6.269, perplexity: 527.696 I0420 16:19:18.308292 132767335823168 metric_logger.py:196] completed step: 8, seconds: 0.058, TFLOP/s/device: 58.662, Tokens/s/device: 8842.222, total_weights: 16384, loss: 6.031, lm_loss: 6.031, perplexity: 416.165 I0420 16:19:18.365830 132767335823168 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 16:19:18.366401 132767335823168 checkpointing.py:798] Waited 0.0005862712860107422 seconds for step 9 to finish before starting checkpointing. I0420 16:19:18.368541 132767335823168 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 16:19:48.151208 132640642221824 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.387159s (commit=36.163973s, array_metadata_write=1.223185s) I0420 16:19:48.152360 132641694586624 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.272 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 46.08980703353882 s) (per-host) I0420 16:19:48.152487 132641694586624 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.820802s. I0420 16:19:48.877319 132641694586624 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701948909136951, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0420 16:19:49.617973 132641694586624 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000255s. I0420 16:19:50.099218 132640125589248 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776701948909136951, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/_CHECKPOINT_METADATA I0420 16:19:50.239254 132641694586624 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 16:19:51.176110 132641694586624 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.175079s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/items I0420 16:19:51.177895 132641694586624 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0/items I0420 16:19:52.229404 132641694586624 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0 I0420 16:19:53.701420 132641694586624 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0`. I0420 16:19:53.702268 132641694586624 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 52.40 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/0 I0420 16:19:53.703813 132641694586624 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 51.372127s. I0420 16:19:53.704033 132640633829120 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 16:19:53.704167 132640633829120 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 16:19:53.704236 132640633829120 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 16:19:53.706882 132640633829120 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 16:19:53.707058 132767335823168 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 16:19:53.707215 132767335823168 checkpoint_manager.py:1452] Waiting for previous save to complete took 35.338670 seconds. If this number is high, consider checkpointing less frequently. I0420 16:19:53.709215 132767335823168 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9 I0420 16:19:53.711321 132767335823168 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9. I0420 16:19:54.411947 132767335823168 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0420 16:19:54.412039 132767335823168 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 16:19:54.437450 132767335823168 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026284s I0420 16:19:54.438157 132767335823168 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 50.351 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030636310577392578 s) (per-host) I0420 16:19:54.438220 132767335823168 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030715s (batch_requests_ready=0.001808s, total_serialization_initiated=0.028286s, others=0.000620s) I0420 16:19:54.438294 132767335823168 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034815s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.034798) I0420 16:19:54.439003 132767335823168 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.73 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9. I0420 16:19:54.439293 132640633829120 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-20 16:39:54.439271 I0420 16:19:54.442239 132767335823168 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 16:19:54.442475 132640100411136 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 16:19:54.442619 132767335823168 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776701958.3685098, 'wait_for_prev_duration_secs': 35.33867049217224, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776701993.7092578, 'checkpointer_blocking_duration_secs': 0.7301304340362549, 'get_old_steps_start_time': 1776701994.4394114, 'get_old_steps_duration_secs': 3.2901763916015625e-05, 'checkpoint_manager_blocking_start_time': 1776701958.3666234, 'checkpoint_manager_blocking_duration_secs': 36.075950622558594} I0420 16:19:54.442739 132767335823168 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 16:19:54.442793 132767335823168 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 16:19:54.676967 132641694586624 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9 I0420 16:19:56.875225 132640092018432 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9/items I0420 16:19:59.923273 132640643880704 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_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9/items/array_metadatas/process_0 I0420 16:20:38.168587 132640612058880 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 40.054936s (commit=38.761877s, array_metadata_write=1.293059s) I0420 16:20:38.169669 132640633829120 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 36.095 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.76210427284241 s) (per-host) I0420 16:20:38.169798 132640633829120 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.730432s. I0420 16:20:40.331595 132640633829120 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000221s. I0420 16:20:40.928034 132640633829120 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389). I0420 16:20:41.879035 132640633829120 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.215332s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9/items I0420 16:20:41.880757 132640633829120 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9/items I0420 16:20:42.434021 132640633829120 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9 I0420 16:20:43.826453 132640633829120 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9`. I0420 16:20:43.827513 132640633829120 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.12 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260420_153543/linen_xpk_feat_nnx_post_train_fixes_20260420_153543_09_pdb_lt_1/checkpoints/9 I0420 16:20:43.828909 132640633829120 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 49.389541s. I0420 16:20:43.829109 132640100411136 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 16:20:43.829222 132640100411136 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 16:20:43.829286 132640100411136 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 16:20:43.831991 132640100411136 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 16:20:43.832157 132767335823168 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 16:20:43.832267 132767335823168 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 16:20:43.833257 132767335823168 metric_logger.py:196] completed step: 9, seconds: 0.058, TFLOP/s/device: 58.105, Tokens/s/device: 8758.275, total_weights: 16384, loss: 5.892, lm_loss: 5.892, perplexity: 361.955 Per train step: Total TFLOPs: 3.40 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Mon Apr 20 16:20:57 UTC 2026 EXIT_CODE=0