2026-04-20 14:54:15.447380: 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 14:54:15.567693 127149860351104 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0420 14:55:10.174420 127149860351104 max_utils.py:800] System Information: Jax Version: 0.9.2 I0420 14:55:10.174537 127149860351104 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0420 14:55:10.174571 127149860351104 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Apr 6 2026 20:48:10 (1775533690) cl/895581894 I0420 14:55:10.174599 127149860351104 train_utils.py:365] WARNING: 'dataset_path' might be pointing your local file system I0420 14:55:10.174690 127149860351104 train.py:806] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0420 14:55:10.698177 127149860351104 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 14:55:10.698357 127149860351104 maxtext_utils.py:1718] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0420 14:55:10.698574 127149860351104 checkpointing.py:688] Setting up checkpoint logger... I0420 14:55:10.698616 127149860351104 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0420 14:55:10.698657 127149860351104 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0420 14:55:10.699221 127149860351104 base_pytree_checkpoint_handler.py:411] 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 0x73a3cb501010>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0420 14:55:13.096673 127149860351104 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0420 14:55:13.096879 127149860351104 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x739d4f73bc20>}, handler_registry=None I0420 14:55:13.097166 127149860351104 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x739d4f73bc20>` 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 14:55:13.097207 127149860351104 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x739d4fb53f80>` 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 14:55:13.097235 127149860351104 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 0x739d4f73bc20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x739d4f73bc20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x739d4fb53f80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x739d4fb53f80>}). I0420 14:55:13.097522 127149860351104 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0420 14:55:13.097575 127149860351104 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x739d4f752fc0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0420 14:55:13.225195 127149860351104 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints I0420 14:55:13.225440 127149860351104 checkpoint_manager.py:921] [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_hns=False, 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), root_directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x739d4f73a180> I0420 14:55:13.225530 127149860351104 checkpointing.py:302] Checkpoint manager created! I0420 14:55:13.646022 127149860351104 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0420 14:55:13.648980 127149860351104 reader.py:262] Creating a tf.data.Dataset reading 8 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0420 14:55:13.703229 127149860351104 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split train, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0420 14:55:13.727170 127149860351104 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 14:55:13.727231 127149860351104 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0420 14:55:14.338008 127149860351104 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0420 14:55:14.338127 127149860351104 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 14:55:14.440868 127149860351104 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 14:55:14.440956 127149860351104 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 14:55:14.455971 127149860351104 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0420 14:55:14.456024 127149860351104 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0420 14:55:14.477946 127149860351104 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 14:55:14.478009 127149860351104 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 14:55:14.493179 127149860351104 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 14:55:14.493233 127149860351104 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 14:55:14.508337 127149860351104 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 14:55:14.508389 127149860351104 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 14:55:14.530887 127149860351104 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0420 14:55:14.530952 127149860351104 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0420 14:55:14.549937 127149860351104 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0420 14:55:14.549992 127149860351104 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0420 14:55:14.744488 127149860351104 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0420 14:55:14.744592 127149860351104 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0420 14:55:16.122783 127149860351104 maxtext_utils.py:1821] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0420 14:55:16.122896 127149860351104 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 14:55:16.122937 127149860351104 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0420 14:55:16.122979 127149860351104 maxtext_utils.py:1821] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0420 14:55:16.123024 127149860351104 maxtext_utils.py:1821] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0420 14:55:16.123070 127149860351104 maxtext_utils.py:1821] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0420 14:55:16.123111 127149860351104 maxtext_utils.py:1821] 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 14:55:16.123151 127149860351104 maxtext_utils.py:1821] 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 14:55:16.123180 127149860351104 maxtext_utils.py:1821] 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 14:55:16.123208 127149860351104 maxtext_utils.py:1821] 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 14:55:16.123245 127149860351104 maxtext_utils.py:1821] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0420 14:55:16.123280 127149860351104 maxtext_utils.py:1821] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0420 14:55:16.562031 127149860351104 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0420 14:55:16.562138 127149860351104 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0420 14:55:16.576456 127149860351104 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0420 14:55:16.576510 127149860351104 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0420 14:55:24.265355 127149860351104 max_utils.py:791] Total memory size: 3.6 GB, Output size: 1.5 GB, Temp size: 2.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0420 14:55:24.266501 127149860351104 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0420 14:55:24.266744 127149860351104 metric_logger.py:301] number parameters: 1.104 billion I0420 14:55:35.309817 127149860351104 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0420 14:55:35.441994 127149860351104 checkpointing.py:798] Waited 0.13216495513916016 seconds for step 0 to finish before starting checkpointing. I0420 14:55:35.442596 127149860351104 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 14:55:35.442810 127149860351104 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0420 14:55:35.443337 127149860351104 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0. I0420 14:55:35.551715 127149860351104 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0420 14:55:35.648483 127149860351104 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0420 14:55:35.648584 127149860351104 replica_slices.py:410] 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 14:55:35.654911 127034944652864 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 14:55:36.477367 127034934167104 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 14:55:36.487725 127149860351104 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.840085s I0420 14:55:36.488272 127149860351104 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 13.198 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 935 milliseconds) (per-host) I0420 14:55:36.488325 127149860351104 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.935148s (batch_requests_ready=0.090618s, total_serialization_initiated=0.844122s, others=0.000408s) I0420 14:55:36.488385 127149860351104 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.936098s (all_items=0.000024s, per_item={'items': '0.00002384'}, temp_paths=0.936074) I0420 14:55:36.489364 127034864961088 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 14:55:36.489470 127149860351104 async_checkpointer.py:561] Finished blocking save. Time taken: 1.046617s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0. I0420 14:55:36.489669 127149860351104 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0420 14:55:36.489890 127149860351104 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776696935.4425814, 'wait_for_prev_duration_secs': 4.601478576660156e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776696935.4428318, 'checkpointer_blocking_duration_secs': 1.0467548370361328, 'get_old_steps_start_time': 1776696936.4896035, 'get_old_steps_duration_secs': 3.0994415283203125e-05, 'checkpoint_manager_blocking_start_time': 1776696935.4424748, 'checkpoint_manager_blocking_duration_secs': 1.047391414642334} I0420 14:55:36.489968 127149860351104 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0420 14:55:36.490012 127149860351104 max_utils.py:750] Memstats: After params initialized: I0420 14:55:36.490118 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0420 14:55:36.490148 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0420 14:55:36.490172 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0420 14:55:36.490190 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0420 14:55:36.490207 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0420 14:55:36.490225 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0420 14:55:36.490241 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0420 14:55:36.490257 127149860351104 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0420 14:55:36.600551 127034885932608 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776696936318264258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776696936318264258, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 14:55:36.600811 127034854475328 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 14:55:36.960202 1075753 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0420 14:55:36.960287 1075753 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0420 14:55:36.997212 127149860351104 metric_logger.py:196] completed step: 0, seconds: 11.042, TFLOP/s/device: 1.231, Tokens/s/device: 185.475, total_weights: 13328, loss: 10.828, lm_loss: 10.828, perplexity: 50404.086 I0420 14:55:36.998340 127149860351104 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/tensorboard/' I0420 14:55:37.103494 127149860351104 metric_logger.py:196] completed step: 1, seconds: 1.676, TFLOP/s/device: 8.107, Tokens/s/device: 1222.045, total_weights: 12332, loss: 10.828, lm_loss: 10.828, perplexity: 50391.758 I0420 14:55:37.220919 127149860351104 metric_logger.py:196] completed step: 2, seconds: 0.021, TFLOP/s/device: 655.591, Tokens/s/device: 98817.853, total_weights: 15161, loss: 9.823, lm_loss: 9.823, perplexity: 18444.867 I0420 14:55:37.338263 127149860351104 metric_logger.py:196] completed step: 3, seconds: 0.107, TFLOP/s/device: 127.304, Tokens/s/device: 19188.607, total_weights: 13327, loss: 9.405, lm_loss: 9.405, perplexity: 12149.331 I0420 14:55:37.720062 127034944652864 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0420 14:55:57.642687 127149860351104 metric_logger.py:196] completed step: 4, seconds: 0.117, TFLOP/s/device: 115.872, Tokens/s/device: 17465.461, total_weights: 11939, loss: 9.045, lm_loss: 9.045, perplexity: 8471.842 I0420 14:55:57.653549 127149860351104 metric_logger.py:196] completed step: 5, seconds: 0.117, TFLOP/s/device: 115.814, Tokens/s/device: 17456.678, total_weights: 15502, loss: 8.953, lm_loss: 8.953, perplexity: 7728.546 I0420 14:55:57.769984 127149860351104 metric_logger.py:196] completed step: 6, seconds: 20.305, TFLOP/s/device: 0.669, Tokens/s/device: 100.864, total_weights: 13864, loss: 8.775, lm_loss: 8.775, perplexity: 6468.536 I0420 14:55:57.887445 127149860351104 metric_logger.py:196] completed step: 7, seconds: 0.007, TFLOP/s/device: 1984.102, Tokens/s/device: 299065.421, total_weights: 12988, loss: 8.688, lm_loss: 8.688, perplexity: 5928.887 I0420 14:55:58.004882 127149860351104 metric_logger.py:196] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.696, Tokens/s/device: 17438.989, total_weights: 13820, loss: 8.740, lm_loss: 8.740, perplexity: 6250.123 I0420 14:55:58.121335 127149860351104 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0420 14:55:58.122064 127149860351104 checkpointing.py:798] Waited 0.0007512569427490234 seconds for step 9 to finish before starting checkpointing. I0420 14:55:58.122366 127149860351104 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 14:56:02.624927 127034923681344 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 25.662679s (commit=25.219043s, array_metadata_write=0.443636s) I0420 14:56:02.626333 127034864961088 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 466.766 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 27 seconds) (per-host) I0420 14:56:02.626455 127034864961088 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.136938s. I0420 14:56:02.864475 127034864961088 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776696936318264258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 14:56:03.043297 127034864961088 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0420 14:56:03.252993 127034885932608 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776696936318264258, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0420 14:56:03.467176 127034864961088 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 14:56:03.467874 127034864961088 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.566273s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 14:56:03.468509 127034864961088 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0/items I0420 14:56:03.714133 127034864961088 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 14:56:04.405307 127034864961088 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0`. I0420 14:56:04.406076 127034864961088 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 28.963229s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/0 I0420 14:56:04.406157 127034864961088 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 27.916642s. I0420 14:56:04.406382 127034854475328 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 14:56:04.406504 127034854475328 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 14:56:04.406561 127034854475328 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0420 14:56:04.406608 127034854475328 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0420 14:56:04.406712 127149860351104 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0420 14:56:04.406786 127149860351104 checkpoint_manager.py:1441] Waiting for previous save to complete took 6.284443 seconds. If this number is high, consider checkpointing less frequently. I0420 14:56:04.407557 127149860351104 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0420 14:56:04.407870 127149860351104 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9. I0420 14:56:04.575433 127149860351104 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0420 14:56:04.575541 127149860351104 replica_slices.py:410] 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 14:56:04.578223 127034854475328 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 14:56:05.254849 127034676217408 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 14:56:08.645667 127149860351104 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 4.070717s I0420 14:56:08.646267 127149860351104 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.967 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 4 seconds) (per-host) I0420 14:56:08.646324 127149860351104 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 4.159362s (batch_requests_ready=0.084849s, total_serialization_initiated=4.074043s, others=0.000470s) I0420 14:56:08.646389 127149860351104 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 4.159940s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=4.159924) I0420 14:56:08.651913 127034864961088 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0420 14:56:08.651999 127149860351104 async_checkpointer.py:561] Finished blocking save. Time taken: 4.244396s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9. I0420 14:56:08.652229 127149860351104 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0420 14:56:08.652415 127034831406656 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0420 14:56:08.652536 127149860351104 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776696958.1223161, 'wait_for_prev_duration_secs': 6.284442901611328, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776696964.4075835, 'checkpointer_blocking_duration_secs': 4.24454665184021, 'get_old_steps_start_time': 1776696968.6521542, 'get_old_steps_duration_secs': 3.409385681152344e-05, 'checkpoint_manager_blocking_start_time': 1776696958.1222649, 'checkpoint_manager_blocking_duration_secs': 10.530246019363403} I0420 14:56:08.652765 127149860351104 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0420 14:56:08.652803 127149860351104 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0420 14:56:09.400227 127034944652864 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0420 14:56:44.788649 127034676217408 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.137637s (commit=35.678849s, array_metadata_write=0.458788s) I0420 14:56:44.789870 127034864961088 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 313.545 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0420 14:56:44.789931 127034864961088 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.137867s. I0420 14:56:45.209737 127034864961088 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0420 14:56:45.623035 127034864961088 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0420 14:56:45.623772 127034864961088 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.569604s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 14:56:45.624547 127034864961088 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9/items I0420 14:56:45.863951 127034864961088 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 14:56:46.547718 127034864961088 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9`. I0420 14:56:46.548451 127034864961088 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.140853s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_trainstate_and_training_loop_20260420_142345/linen_feat_nnx_trainstate_and_training_loop_20260420_142345_10_shardy_false/checkpoints/9 I0420 14:56:46.548524 127034864961088 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.896461s. I0420 14:56:46.548698 127034831406656 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0420 14:56:46.548817 127034831406656 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0420 14:56:46.548886 127034831406656 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0420 14:56:46.548931 127034831406656 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0420 14:56:46.549122 127149860351104 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0420 14:56:46.549276 127149860351104 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0420 14:56:46.551176 127149860351104 metric_logger.py:196] completed step: 9, seconds: 0.117, TFLOP/s/device: 116.044, Tokens/s/device: 17491.417, total_weights: 12300, loss: 8.710, lm_loss: 8.710, perplexity: 6065.008 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops