2026-04-16 18:20:52.304725: 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 18:20:52.420080 124577722178688 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0416 18:21:45.210633 124577722178688 max_utils.py:800] System Information: Jax Version: 0.9.2 I0416 18:21:45.210740 124577722178688 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0416 18:21:45.210776 124577722178688 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 18:21:45.210802 124577722178688 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system I0416 18:21:45.210892 124577722178688 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0416 18:21:45.306782 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:21:45.710394 124577722178688 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:21:45.710562 124577722178688 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0416 18:21:45.710800 124577722178688 checkpointing.py:688] Setting up checkpoint logger... I0416 18:21:45.710849 124577722178688 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0416 18:21:45.710888 124577722178688 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0416 18:21:45.711619 124577722178688 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 0x714cec9fcc20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0416 18:21:48.102331 124577722178688 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0416 18:21:48.102503 124577722178688 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 0x714670f25b20>}, handler_registry=None I0416 18:21:48.102768 124577722178688 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x714670f25b20>` 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 18:21:48.102810 124577722178688 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714670138800>` 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 18:21:48.102837 124577722178688 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 0x714670f25b20>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x714670f25b20>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714670138800>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x714670138800>}). I0416 18:21:48.103133 124577722178688 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0416 18:21:48.103181 124577722178688 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x714670135ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0416 18:21:48.235471 124577722178688 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_03_dropout/checkpoints I0416 18:21:48.235695 124577722178688 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_03_dropout/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7147e73a8920> I0416 18:21:48.235772 124577722178688 checkpointing.py:302] Checkpoint manager created! I0416 18:21:48.316151 124577722178688 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0416 18:21:48.319092 124577722178688 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 18:21:48.372419 124577722178688 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 18:21:48.403521 124577722178688 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:21:48.403592 124577722178688 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0416 18:21:49.395596 124577722178688 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_norm_length', 'activation_embed'). I0416 18:21:49.395692 124577722178688 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0416 18:21:49.500094 124577722178688 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:21:49.500172 124577722178688 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0416 18:21:49.515039 124577722178688 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,128,2048]........................................ ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0416 18:21:49.515098 124577722178688 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,128,2048]........................................ ('fsdp', None, None). I0416 18:21:49.537404 124577722178688 attentions.py:1154] attentions/query Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:21:49.537459 124577722178688 attentions.py:1154] attentions/query Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0416 18:21:49.552368 124577722178688 attentions.py:1155] attentions/key Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:21:49.552417 124577722178688 attentions.py:1155] attentions/key Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0416 18:21:49.567344 124577722178688 attentions.py:1156] attentions/value Logical: bfloat16[8,128,16,128]...................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0416 18:21:49.567393 124577722178688 attentions.py:1156] attentions/value Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0416 18:21:49.589331 124577722178688 attentions.py:1197] attentions/out Logical: bfloat16[8,128,16,128]...................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0416 18:21:49.589385 124577722178688 attentions.py:1197] attentions/out Physical: bfloat16[8,128,16,128]...................................... ('fsdp', None, None, None). I0416 18:21:49.612473 124577722178688 linears.py:525] linears/x Logical: bfloat16[8,128,7168]........................................ ('activation_batch', 'activation_length', 'activation_mlp'). I0416 18:21:49.612528 124577722178688 linears.py:525] linears/x Physical: bfloat16[8,128,7168]........................................ ('fsdp', None, None). I0416 18:21:49.815653 124577722178688 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0416 18:21:49.815747 124577722178688 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. W0416 18:21:49.865407 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. [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 18:21:50.266812 124577722178688 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0416 18:21:50.266918 124577722178688 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 18:21:50.266960 124577722178688 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 18:21:50.267004 124577722178688 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 18:21:50.267058 124577722178688 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 18:21:50.267090 124577722178688 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 18:21:50.267130 124577722178688 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 18:21:50.267174 124577722178688 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 18:21:50.267203 124577722178688 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 18:21:50.267230 124577722178688 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 18:21:50.267264 124577722178688 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0416 18:21:50.267299 124577722178688 maxtext_utils.py:1796] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0416 18:21:50.730787 124577722178688 train.py:157] train/xent Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:21:50.730871 124577722178688 train.py:157] train/xent Physical: float32[8,128].............................................. ('fsdp', None). I0416 18:21:50.745620 124577722178688 train.py:164] train/z_loss Logical: float32[8,128].............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0416 18:21:50.745669 124577722178688 train.py:164] train/z_loss Physical: float32[8,128].............................................. ('fsdp', None). W0416 18:21:51.010842 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:21:51.101550 124577722178688 max_utils.py:791] Total memory size: 2.5 GB, Output size: 1.5 GB, Temp size: 1.0 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0416 18:21:51.102005 124577722178688 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0416 18:21:51.102158 124577722178688 metric_logger.py:289] number parameters: 1.104 billion W0416 18:21:54.128064 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:21:54.232685 124577722178688 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0416 18:21:54.298568 124577722178688 checkpointing.py:798] Waited 0.06587624549865723 seconds for step 0 to finish before starting checkpointing. I0416 18:21:54.298968 124577722178688 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:21:54.299133 124577722178688 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0416 18:21:54.299492 124577722178688 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_03_dropout/checkpoints/0. I0416 18:21:54.384036 124577722178688 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0416 18:21:54.470971 124464459155008 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_03_dropout/checkpoints/0 I0416 18:21:54.484365 124577722178688 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:21:54.484451 124577722178688 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 W0416 18:21:54.699342 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.716827 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.731512 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.746949 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.757763 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.763706 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.768920 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:54.773645 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:21:55.088982 124577722178688 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.605028s I0416 18:21:55.089419 124577722178688 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 17.529 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 704 milliseconds) (per-host) I0416 18:21:55.089471 124577722178688 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.704111s (batch_requests_ready=0.095309s, total_serialization_initiated=0.608478s, others=0.000324s) I0416 18:21:55.089531 124577722178688 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.705034s (all_items=0.000021s, per_item={'items': '0.00002122'}, temp_paths=0.705013) I0416 18:21:55.090259 124464387851840 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:21:55.090350 124577722178688 async_checkpointer.py:561] Finished blocking save. Time taken: 0.791177s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/0. I0416 18:21:55.090517 124577722178688 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:21:55.090685 124464469640768 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:21:55.090788 124577722178688 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_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776363714.2989557, 'wait_for_prev_duration_secs': 4.0531158447265625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776363714.2991552, 'checkpointer_blocking_duration_secs': 0.7912919521331787, 'get_old_steps_start_time': 1776363715.0904646, 'get_old_steps_duration_secs': 2.002716064453125e-05, 'checkpoint_manager_blocking_start_time': 1776363714.298874, 'checkpoint_manager_blocking_duration_secs': 0.7918868064880371} I0416 18:21:55.090893 124577722178688 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0416 18:21:55.090932 124577722178688 max_utils.py:750] Memstats: After params initialized: I0416 18:21:55.091008 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_0(process=0,(0,0,0,0)) I0416 18:21:55.091034 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_1(process=0,(1,0,0,0)) I0416 18:21:55.091069 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_2(process=0,(0,1,0,0)) I0416 18:21:55.091089 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_3(process=0,(1,1,0,0)) I0416 18:21:55.091108 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_4(process=0,(0,2,0,0)) I0416 18:21:55.091125 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_5(process=0,(1,2,0,0)) I0416 18:21:55.091143 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_6(process=0,(0,3,0,0)) I0416 18:21:55.091160 124577722178688 max_utils.py:756] Using (GB) 1.56 / 31.25 (4.992000%) on TPU_7(process=0,(1,3,0,0)) W0416 18:21:55.096910 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:55.102640 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0416 18:21:55.107419 3973927 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0416 18:21:55.198417 124464448669248 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_03_dropout/checkpoints/0/items I0416 18:21:55.422818 124577722178688 metric_logger.py:185] completed step: 0, seconds: 3.130, TFLOP/s/device: 0.256, Tokens/s/device: 40.894, total_weights: 963, loss: 10.834 I0416 18:21:55.423672 124577722178688 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_03_dropout/tensorboard/' I0416 18:21:55.484291 124577722178688 metric_logger.py:185] completed step: 1, seconds: 1.188, TFLOP/s/device: 0.674, Tokens/s/device: 107.712, total_weights: 945, loss: 10.747 I0416 18:21:55.548196 124577722178688 metric_logger.py:185] completed step: 2, seconds: 0.009, TFLOP/s/device: 89.594, Tokens/s/device: 14319.275, total_weights: 962, loss: 10.246 I0416 18:21:55.569535 124464377366080 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363715112502221, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776363715112502221, "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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 18:21:55.611834 124577722178688 metric_logger.py:185] completed step: 3, seconds: 0.062, TFLOP/s/device: 12.857, Tokens/s/device: 2054.839, total_weights: 975, loss: 9.829 I0416 18:21:55.877866 3976538 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0416 18:21:55.877933 3976538 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0416 18:21:56.652055 124464408823360 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_03_dropout/checkpoints/0/items/array_metadatas/process_0 I0416 18:22:06.182255 124577722178688 metric_logger.py:185] completed step: 4, seconds: 0.064, TFLOP/s/device: 12.480, Tokens/s/device: 1994.670, total_weights: 962, loss: 9.393 I0416 18:22:06.192623 124577722178688 metric_logger.py:185] completed step: 5, seconds: 0.064, TFLOP/s/device: 12.527, Tokens/s/device: 2002.065, total_weights: 958, loss: 9.248 I0416 18:22:06.254792 124577722178688 metric_logger.py:185] completed step: 6, seconds: 10.571, TFLOP/s/device: 0.076, Tokens/s/device: 12.108, total_weights: 930, loss: 9.424 I0416 18:22:06.318422 124577722178688 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 127.023, Tokens/s/device: 20301.348, total_weights: 935, loss: 9.317 I0416 18:22:06.381994 124577722178688 metric_logger.py:185] completed step: 8, seconds: 0.063, TFLOP/s/device: 12.740, Tokens/s/device: 2036.142, total_weights: 930, loss: 9.478 I0416 18:22:06.445418 124577722178688 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0416 18:22:06.446138 124577722178688 checkpointing.py:798] Waited 0.0007476806640625 seconds for step 9 to finish before starting checkpointing. I0416 18:22:06.446409 124577722178688 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:22:23.368407 124464398337600 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 27.488995s (commit=27.063060s, array_metadata_write=0.425935s) I0416 18:22:23.369423 124464387851840 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 435.991 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 28 seconds) (per-host) I0416 18:22:23.369481 124464387851840 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 28.279088s. I0416 18:22:23.647227 124464387851840 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363715112502221, '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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 18:22:23.826781 124464387851840 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 18:22:24.042289 124464377366080 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363715112502221, '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_03_dropout/checkpoints/0/_CHECKPOINT_METADATA I0416 18:22:24.263926 124464387851840 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:22:24.264108 124464387851840 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.575136s. 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_03_dropout/checkpoints/0/items I0416 18:22:24.264720 124464387851840 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/0/items I0416 18:22:24.511723 124464387851840 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/0 I0416 18:22:25.202654 124464387851840 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_03_dropout/checkpoints/0`. I0416 18:22:25.203282 124464387851840 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 30.904115s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/0 I0416 18:22:25.203356 124464387851840 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 30.112964s. I0416 18:22:25.203547 124464469640768 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:22:25.203658 124464469640768 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:22:25.203722 124464469640768 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:22:25.203769 124464469640768 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0416 18:22:25.203912 124577722178688 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 18:22:25.204031 124577722178688 checkpoint_manager.py:1441] Waiting for previous save to complete took 18.757638 seconds. If this number is high, consider checkpointing less frequently. I0416 18:22:25.205106 124577722178688 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0416 18:22:25.205384 124577722178688 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_03_dropout/checkpoints/9. I0416 18:22:25.381626 124464469640768 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_03_dropout/checkpoints/9 I0416 18:22:25.385735 124577722178688 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0416 18:22:25.385826 124577722178688 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 18:22:26.079368 124464251536960 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_03_dropout/checkpoints/9/items I0416 18:22:34.454747 124577722178688 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.069411s I0416 18:22:34.458799 124577722178688 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.347 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0416 18:22:34.458868 124577722178688 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.159161s (batch_requests_ready=0.082593s, total_serialization_initiated=9.072614s, others=0.003954s) I0416 18:22:34.458941 124577722178688 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.159667s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=9.159650) I0416 18:22:34.461025 124464398337600 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0416 18:22:34.461196 124577722178688 async_checkpointer.py:561] Finished blocking save. Time taken: 9.256047s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/9. I0416 18:22:34.461396 124577722178688 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0416 18:22:34.461530 124464387851840 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0416 18:22:34.461603 124577722178688 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_03_dropout/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776363726.4463646, 'wait_for_prev_duration_secs': 18.757638216018677, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776363745.2051303, 'checkpointer_blocking_duration_secs': 9.256181955337524, 'get_old_steps_start_time': 1776363754.461327, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776363726.4463215, 'checkpoint_manager_blocking_duration_secs': 28.015256643295288} I0416 18:22:34.461709 124577722178688 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0416 18:22:34.461741 124577722178688 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0416 18:22:35.256335 124464438183488 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_03_dropout/checkpoints/9/items/array_metadatas/process_0 I0416 18:23:09.924835 124464251536960 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.465326s (commit=35.031826s, array_metadata_write=0.433500s) I0416 18:23:09.925864 124464398337600 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 283.170 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0416 18:23:09.925971 124464398337600 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.464733s. I0416 18:23:10.331687 124464398337600 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 18:23:10.789701 124464398337600 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0416 18:23:10.789881 124464398337600 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.606074s. 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_03_dropout/checkpoints/9/items I0416 18:23:10.790550 124464398337600 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/9/items I0416 18:23:11.038780 124464398337600 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/9 I0416 18:23:11.729896 124464398337600 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_03_dropout/checkpoints/9`. I0416 18:23:11.730530 124464398337600 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.525388s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_03_dropout/checkpoints/9 I0416 18:23:11.730617 124464398337600 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.269380s. I0416 18:23:11.730739 124464387851840 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0416 18:23:11.730787 124464387851840 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0416 18:23:11.730837 124464387851840 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0416 18:23:11.730878 124464387851840 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0416 18:23:11.731131 124577722178688 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 18:23:11.731259 124577722178688 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0416 18:23:11.731830 124577722178688 metric_logger.py:185] completed step: 9, seconds: 0.063, TFLOP/s/device: 12.642, Tokens/s/device: 2020.457, total_weights: 948, loss: 9.141 Per train step: Total TFLOPs: 0.80 split as 99.60% learnable weight flops and 0.40% attention flops