2026-04-16 18:48:53.345973: 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) I0416 18:48:53.460873 133823608179840 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 18:49:48.689648 133823608179840 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 18:49:48.689755 133823608179840 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 18:49:48.689791 133823608179840 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 I0416 18:49:48.689816 133823608179840 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 18:49:48.689899 133823608179840 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0416 18:49:49.199733 133823608179840 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:49:49.199906 133823608179840 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:49:49.200135 133823608179840 checkpointing.py:688] Setting up checkpoint logger... I0416 18:49:49.200176 133823608179840 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 18:49:49.200208 133823608179840 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 18:49:49.200980 133823608179840 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 0x79b5a65dbe90>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 18:49:51.582737 133823608179840 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 18:49:51.582901 133823608179840 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 0x79b0a0f66780>}, handler_registry=None I0416 18:49:51.583154 133823608179840 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b0a0f66780>` 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`. I0416 18:49:51.583195 133823608179840 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79af27d42120>` 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`. I0416 18:49:51.583229 133823608179840 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 0x79b0a0f66780>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79b0a0f66780>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79af27d42120>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79af27d42120>}). I0416 18:49:51.583505 133823608179840 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 18:49:51.583553 133823608179840 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79af27d31ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 18:49:51.705937 133823608179840 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints I0416 18:49:51.706175 133823608179840 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_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79af27d41130> I0416 18:49:51.706252 133823608179840 checkpointing.py:302] Checkpoint manager created! I0416 18:49:51.786587 133823608179840 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0416 18:49:51.789531 133823608179840 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. I0416 18:49:51.844093 133823608179840 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 I0416 18:49:51.872172 133823608179840 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:49:51.872234 133823608179840 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:49:52.837277 133823608179840 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 18:49:52.837371 133823608179840 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:49:52.940279 133823608179840 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:49:52.940356 133823608179840 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:49:52.955335 133823608179840 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:49:52.955384 133823608179840 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0416 18:49:52.976815 133823608179840 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:49:52.976872 133823608179840 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:49:52.992069 133823608179840 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:49:52.992121 133823608179840 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:49:53.007134 133823608179840 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:49:53.007182 133823608179840 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:49:53.030670 133823608179840 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 18:49:53.030727 133823608179840 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0416 18:49:53.049458 133823608179840 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0416 18:49:53.049510 133823608179840 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0416 18:49:53.245617 133823608179840 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 18:49:53.245710 133823608179840 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 I0416 18:49:54.647924 133823608179840 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 18:49:54.648058 133823608179840 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 18:49:54.648103 133823608179840 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0416 18:49:54.648149 133823608179840 maxtext_utils.py:1796] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0416 18:49:54.648189 133823608179840 maxtext_utils.py:1796] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 18:49:54.648222 133823608179840 maxtext_utils.py:1796] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0416 18:49:54.648262 133823608179840 maxtext_utils.py:1796] 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) I0416 18:49:54.648305 133823608179840 maxtext_utils.py:1796] 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') I0416 18:49:54.648334 133823608179840 maxtext_utils.py:1796] 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) I0416 18:49:54.648363 133823608179840 maxtext_utils.py:1796] 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) I0416 18:49:54.648397 133823608179840 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 18:49:54.648433 133823608179840 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 18:49:55.087506 133823608179840 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:49:55.087593 133823608179840 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0416 18:49:55.102233 133823608179840 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:49:55.102285 133823608179840 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0416 18:49:55.390771 4003052 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:49:55.504887 133823608179840 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. I0416 18:49:55.505379 133823608179840 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 18:49:55.505531 133823608179840 metric_logger.py:289] number parameters: 1.104 billion W0416 18:49:58.544893 4003052 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:49:58.672921 133823608179840 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 18:49:58.794461 133823608179840 checkpointing.py:798] Waited 0.12152528762817383 seconds for step 0 to finish before starting checkpointing. I0416 18:49:58.794908 133823608179840 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:49:58.795073 133823608179840 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 18:49:58.795447 133823608179840 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0. I0416 18:49:58.876829 133823608179840 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 18:49:58.961561 133708742919744 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 18:49:58.966527 133823608179840 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:49:58.966621 133823608179840 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 I0416 18:49:59.765282 133823608179840 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.799150s I0416 18:49:59.765482 133708732433984 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 18:49:59.766256 133823608179840 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 13.891 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 888 milliseconds) (per-host) I0416 18:49:59.766340 133823608179840 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.888564s (batch_requests_ready=0.084984s, total_serialization_initiated=0.802898s, others=0.000681s) I0416 18:49:59.766419 133823608179840 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.889116s (all_items=0.000019s, per_item={'items': '0.00001907'}, temp_paths=0.889097) I0416 18:49:59.767606 133708663227968 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:49:59.767752 133823608179840 async_checkpointer.py:561] Finished blocking save. Time taken: 0.972632s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0. I0416 18:49:59.768011 133823608179840 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:49:59.768262 133823608179840 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776365398.7948952, 'wait_for_prev_duration_secs': 4.100799560546875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776365398.7950976, 'checkpointer_blocking_duration_secs': 0.9728107452392578, 'get_old_steps_start_time': 1776365399.7679327, 'get_old_steps_duration_secs': 3.3855438232421875e-05, 'checkpoint_manager_blocking_start_time': 1776365398.7948115, 'checkpoint_manager_blocking_duration_secs': 0.973419189453125} I0416 18:49:59.768351 133823608179840 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 18:49:59.768419 133823608179840 max_utils.py:750] Memstats: After params initialized: I0416 18:49:59.768517 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0416 18:49:59.768543 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0416 18:49:59.768563 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0416 18:49:59.768581 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0416 18:49:59.768598 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0416 18:49:59.768615 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0416 18:49:59.768633 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0416 18:49:59.768650 133823608179840 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) I0416 18:49:59.938739 133708684199488 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365399666065354, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776365399666065354, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 18:49:59.939068 133708753405504 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:50:00.203231 133823608179840 metric_logger.py:185] completed step: 0, seconds: 3.167, TFLOP/s/device: 4.290, Tokens/s/device: 646.697, total_weights: 13328, loss: 10.828 I0416 18:50:00.204412 133823608179840 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/tensorboard/' I0416 18:50:00.256319 4006055 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 18:50:00.256381 4006055 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 18:50:00.318548 133823608179840 metric_logger.py:185] completed step: 1, seconds: 1.527, TFLOP/s/device: 8.896, Tokens/s/device: 1340.848, total_weights: 12332, loss: 10.828 I0416 18:50:00.436459 133823608179840 metric_logger.py:185] completed step: 2, seconds: 0.011, TFLOP/s/device: 1186.234, Tokens/s/device: 178802.165, total_weights: 15161, loss: 9.823 I0416 18:50:00.554129 133823608179840 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 115.981, Tokens/s/device: 17481.861, total_weights: 13327, loss: 9.405 I0416 18:50:01.232711 133708721948224 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items/array_metadatas/process_0 I0416 18:50:19.865703 133823608179840 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 115.323, Tokens/s/device: 17382.743, total_weights: 11939, loss: 9.045 I0416 18:50:19.876551 133823608179840 metric_logger.py:185] completed step: 5, seconds: 0.117, TFLOP/s/device: 115.770, Tokens/s/device: 17450.133, total_weights: 15502, loss: 8.953 I0416 18:50:19.992801 133823608179840 metric_logger.py:185] completed step: 6, seconds: 19.311, TFLOP/s/device: 0.704, Tokens/s/device: 106.052, total_weights: 13864, loss: 8.775 I0416 18:50:20.110485 133823608179840 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2100.020, Tokens/s/device: 316537.867, total_weights: 12988, loss: 8.688 I0416 18:50:20.227986 133823608179840 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 116.250, Tokens/s/device: 17522.395, total_weights: 13820, loss: 8.740 I0416 18:50:20.345429 133823608179840 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 18:50:20.346213 133823608179840 checkpointing.py:798] Waited 0.0008153915405273438 seconds for step 9 to finish before starting checkpointing. I0416 18:50:20.346493 133823608179840 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:50:26.437915 133708742919744 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 26.188361s (commit=25.663841s, array_metadata_write=0.524521s) I0416 18:50:26.439843 133708663227968 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 458.487 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 27 seconds) (per-host) I0416 18:50:26.440015 133708663227968 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 26.672197s. I0416 18:50:26.676154 133708663227968 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365399666065354, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 18:50:26.866069 133708663227968 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 18:50:27.079227 133708684199488 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776365399666065354, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/_CHECKPOINT_METADATA I0416 18:50:27.277634 133708663227968 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:50:27.278516 133708663227968 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.560110s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 18:50:27.279494 133708663227968 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0/items I0416 18:50:27.525082 133708663227968 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 18:50:28.177366 133708663227968 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0`. I0416 18:50:28.178282 133708663227968 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 29.383170s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/0 I0416 18:50:28.178372 133708663227968 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 28.410561s. I0416 18:50:28.178611 133708753405504 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:50:28.178765 133708753405504 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:50:28.178852 133708753405504 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:50:28.178907 133708753405504 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 18:50:28.179102 133823608179840 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. W0416 18:50:28.179273 133823608179840 checkpoint_manager.py:1441] Waiting for previous save to complete took 7.832769 seconds. If this number is high, consider checkpointing less frequently. I0416 18:50:28.180536 133823608179840 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 18:50:28.180874 133823608179840 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9. I0416 18:50:28.364965 133708753405504 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 18:50:28.370059 133823608179840 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:50:28.370158 133823608179840 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 I0416 18:50:29.031160 133708531107392 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 18:50:38.084910 133823608179840 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.715277s I0416 18:50:38.085483 133823608179840 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.258 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0416 18:50:38.085540 133823608179840 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.810092s (batch_requests_ready=0.091129s, total_serialization_initiated=9.718529s, others=0.000434s) I0416 18:50:38.085617 133823608179840 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.810671s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=9.810652) I0416 18:50:38.090835 133708753405504 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:50:38.090941 133823608179840 async_checkpointer.py:561] Finished blocking save. Time taken: 9.910352s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9. I0416 18:50:38.091158 133823608179840 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:50:38.091361 133708663227968 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:50:38.091475 133823608179840 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776365420.3464482, 'wait_for_prev_duration_secs': 7.832769155502319, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776365428.1805627, 'checkpointer_blocking_duration_secs': 9.910501956939697, 'get_old_steps_start_time': 1776365438.0910797, 'get_old_steps_duration_secs': 3.743171691894531e-05, 'checkpoint_manager_blocking_start_time': 1776365420.3464003, 'checkpoint_manager_blocking_duration_secs': 17.74504780769348} I0416 18:50:38.091608 133823608179840 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 18:50:38.091643 133823608179840 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:50:38.814483 133708721948224 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items/array_metadatas/process_0 I0416 18:51:07.537548 133708531107392 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 29.447510s (commit=29.000322s, array_metadata_write=0.447189s) I0416 18:51:07.538690 133708753405504 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 321.848 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 39 seconds) (per-host) I0416 18:51:07.538748 133708753405504 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.447761s. I0416 18:51:07.934784 133708753405504 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0416 18:51:08.368590 133708753405504 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:51:08.369156 133708753405504 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.564705s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 18:51:08.369794 133708753405504 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9/items I0416 18:51:08.630858 133708753405504 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 18:51:09.316349 133708753405504 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9`. I0416 18:51:09.316972 133708753405504 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 41.136397s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_10_shardy_false/checkpoints/9 I0416 18:51:09.317065 133708753405504 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.226079s. I0416 18:51:09.317262 133708663227968 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:51:09.317370 133708663227968 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:51:09.317448 133708663227968 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:51:09.317494 133708663227968 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 18:51:09.317914 133823608179840 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. I0416 18:51:09.318063 133823608179840 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:51:09.319203 133823608179840 metric_logger.py:185] completed step: 9, seconds: 0.118, TFLOP/s/device: 115.434, Tokens/s/device: 17399.431, total_weights: 12300, loss: 8.710 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops