2026-04-15 23:25:01.792798: 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) I0415 23:25:01.910377 134049108950144 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0415 23:25:54.750816 134049108950144 max_utils.py:800] System Information: Jax Version: 0.9.2 I0415 23:25:54.750921 134049108950144 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0415 23:25:54.750958 134049108950144 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 I0415 23:25:54.750985 134049108950144 train_utils.py:334] WARNING: 'dataset_path' might be pointing your local file system I0415 23:25:54.751086 134049108950144 train.py:715] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. W0415 23:25:54.842168 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:25:55.238878 134049108950144 maxtext_utils.py:1520] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0415 23:25:55.239164 134049108950144 checkpointing.py:677] Setting up checkpoint logger... I0415 23:25:55.239210 134049108950144 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0415 23:25:55.239250 134049108950144 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0415 23:25:55.239712 134049108950144 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 0x79ea269f8b00>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0415 23:25:57.647923 134049108950144 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0415 23:25:57.648338 134049108950144 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 0x79e521ca4c50>}, handler_registry=None I0415 23:25:57.648895 134049108950144 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79e521ca4c50>` 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`. I0415 23:25:57.648939 134049108950144 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79e3aa736600>` 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`. I0415 23:25:57.648971 134049108950144 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 0x79e521ca4c50>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79e521ca4c50>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79e3aa736600>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79e3aa736600>}). I0415 23:25:57.649794 134049108950144 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0415 23:25:57.649872 134049108950144 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79e3aa7394e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0415 23:25:57.768650 134049108950144 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints I0415 23:25:57.768911 134049108950144 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_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79e56bdf43e0> I0415 23:25:57.769013 134049108950144 checkpointing.py:301] Checkpoint manager created! I0415 23:25:57.853558 134049108950144 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0415 23:25:57.856700 134049108950144 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. I0415 23:25:57.911407 134049108950144 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 I0415 23:25:57.944734 134049108950144 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0415 23:25:57.944801 134049108950144 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0415 23:25:58.478999 134049108950144 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0415 23:25:58.480559 134049108950144 reader.py:262] Creating a tf.data.Dataset reading 2 files located in folders: tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0. I0415 23:25:58.496518 134049108950144 logging_logger.py:49] Constructing tf.data.Dataset __local_c4_builder for split validation, from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0 I0415 23:25:58.499117 134049108950144 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2 I0415 23:25:58.499166 134049108950144 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2 I0415 23:25:59.191644 134049108950144 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0415 23:25:59.191745 134049108950144 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:25:59.295700 134049108950144 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:25:59.295780 134049108950144 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:25:59.310734 134049108950144 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0415 23:25:59.310783 134049108950144 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0415 23:25:59.332289 134049108950144 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:25:59.332342 134049108950144 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:25:59.347291 134049108950144 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:25:59.347341 134049108950144 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:25:59.362285 134049108950144 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0415 23:25:59.362334 134049108950144 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:25:59.385359 134049108950144 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0415 23:25:59.385420 134049108950144 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0415 23:25:59.404788 134049108950144 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0415 23:25:59.404842 134049108950144 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0415 23:25:59.590357 134049108950144 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0415 23:25:59.590448 134049108950144 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. W0415 23:25:59.641795 3368867 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 I0415 23:26:00.038708 134049108950144 maxtext_utils.py:1623] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0415 23:26:00.038820 134049108950144 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:26:00.038861 134049108950144 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0415 23:26:00.038905 134049108950144 maxtext_utils.py:1623] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0415 23:26:00.038944 134049108950144 maxtext_utils.py:1623] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:26:00.038976 134049108950144 maxtext_utils.py:1623] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0415 23:26:00.039017 134049108950144 maxtext_utils.py:1623] 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) I0415 23:26:00.039074 134049108950144 maxtext_utils.py:1623] 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') I0415 23:26:00.039108 134049108950144 maxtext_utils.py:1623] 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) I0415 23:26:00.039137 134049108950144 maxtext_utils.py:1623] 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) I0415 23:26:00.039172 134049108950144 maxtext_utils.py:1623] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed', 'vocab') Physical: ('fsdp', None) I0415 23:26:00.039207 134049108950144 maxtext_utils.py:1623] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed') Physical: (None, 'fsdp') I0415 23:26:00.471829 134049108950144 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:26:00.471910 134049108950144 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0415 23:26:00.486230 134049108950144 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0415 23:26:00.486280 134049108950144 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). W0415 23:26:00.782783 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:26:00.902312 134049108950144 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. I0415 23:26:00.902809 134049108950144 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0415 23:26:00.902949 134049108950144 metric_logger.py:289] number parameters: 1.104 billion W0415 23:26:03.990361 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:26:04.124910 134049108950144 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0415 23:26:04.244568 134049108950144 checkpointing.py:776] Waited 0.1196599006652832 seconds for step 0 to finish before starting checkpointing. I0415 23:26:04.245262 134049108950144 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:26:04.245458 134049108950144 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0415 23:26:04.245902 134049108950144 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0. I0415 23:26:04.354521 134049108950144 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0415 23:26:04.449848 134049108950144 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:26:04.449941 134049108950144 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 I0415 23:26:04.458025 133934035764800 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0 W0415 23:26:04.670749 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.685078 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.702040 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.717560 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.727481 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.734034 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.738361 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:04.743692 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:26:05.015623 134049108950144 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.566234s I0415 23:26:05.016080 134049108950144 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 18.692 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 660 milliseconds) (per-host) I0415 23:26:05.016135 134049108950144 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.660321s (batch_requests_ready=0.090102s, total_serialization_initiated=0.569830s, others=0.000389s) I0415 23:26:05.016194 134049108950144 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.661191s (all_items=0.000037s, per_item={'items': '0.00003719'}, temp_paths=0.661154) I0415 23:26:05.016997 133933964461632 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:26:05.017132 134049108950144 async_checkpointer.py:561] Finished blocking save. Time taken: 0.771629s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0. I0415 23:26:05.017281 134049108950144 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:26:05.017493 133934046250560 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:26:05.017614 134049108950144 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776295564.2452426, 'wait_for_prev_duration_secs': 4.935264587402344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776295564.245479, 'checkpointer_blocking_duration_secs': 0.77170729637146, 'get_old_steps_start_time': 1776295565.0172071, 'get_old_steps_duration_secs': 3.600120544433594e-05, 'checkpoint_manager_blocking_start_time': 1776295564.2450674, 'checkpoint_manager_blocking_duration_secs': 0.7725045680999756} I0415 23:26:05.017731 134049108950144 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0415 23:26:05.017790 134049108950144 max_utils.py:750] Memstats: After params initialized: I0415 23:26:05.017865 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0)) I0415 23:26:05.017891 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0)) I0415 23:26:05.017910 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0)) I0415 23:26:05.017934 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0)) I0415 23:26:05.017951 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0)) I0415 23:26:05.017969 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0)) I0415 23:26:05.017984 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0)) I0415 23:26:05.018004 134049108950144 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0)) W0415 23:26:05.023978 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:05.029223 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. W0415 23:26:05.033805 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:26:05.188374 133934025279040 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/items I0415 23:26:05.330555 134049108950144 metric_logger.py:185] completed step: 0, seconds: 3.221, TFLOP/s/device: 4.218, Tokens/s/device: 635.741, total_weights: 13328, loss: 10.828 I0415 23:26:05.331591 134049108950144 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/tensorboard/' I0415 23:26:05.336558 133933945587264 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295565099628013, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776295565099628013, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0415 23:26:05.446637 134049108950144 metric_logger.py:185] completed step: 1, seconds: 1.204, TFLOP/s/device: 11.288, Tokens/s/device: 1701.406, total_weights: 12332, loss: 10.828 I0415 23:26:05.563872 134049108950144 metric_logger.py:185] completed step: 2, seconds: 0.010, TFLOP/s/device: 1376.888, Tokens/s/device: 207539.522, total_weights: 15161, loss: 9.823 I0415 23:26:06.186439 3372190 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0415 23:26:06.186490 3372190 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0415 23:26:07.046288 133933985433152 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/items/array_metadatas/process_0 W0415 23:26:12.386200 3368867 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions. I0415 23:26:12.471657 134049108950144 train.py:632] Completed eval step 0 I0415 23:26:12.513472 134049108950144 train.py:632] Completed eval step 1 I0415 23:26:12.518294 134049108950144 metric_logger.py:204] eval metrics after step: 4, loss=8.979, total_weights=26703.0 I0415 23:26:12.518628 134049108950144 metric_logger.py:185] completed step: 3, seconds: 0.117, TFLOP/s/device: 116.615, Tokens/s/device: 17577.438, total_weights: 13327, loss: 9.405 I0415 23:26:30.841782 133933974947392 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 24.654499s (commit=24.218391s, array_metadata_write=0.436108s) I0415 23:26:30.843080 133933964461632 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 477.090 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 26 seconds) (per-host) I0415 23:26:30.843154 133933964461632 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 25.826046s. I0415 23:26:31.081598 133933964461632 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295565099628013, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0415 23:26:31.270292 133933964461632 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:26:31.468016 133933945587264 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776295565099628013, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/_CHECKPOINT_METADATA I0415 23:26:31.704289 133933964461632 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:26:31.704469 133933964461632 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.578497s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/items I0415 23:26:31.705592 133933964461632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0/items I0415 23:26:31.947017 133933964461632 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0 I0415 23:26:32.014353 134049108950144 metric_logger.py:185] completed step: 4, seconds: 0.117, TFLOP/s/device: 115.670, Tokens/s/device: 17434.981, total_weights: 11939, loss: 9.045 I0415 23:26:32.024335 134049108950144 metric_logger.py:185] completed step: 5, seconds: 6.954, TFLOP/s/device: 1.954, Tokens/s/device: 294.517, total_weights: 15502, loss: 8.953 I0415 23:26:32.141123 134049108950144 metric_logger.py:185] completed step: 6, seconds: 19.497, TFLOP/s/device: 0.697, Tokens/s/device: 105.041, total_weights: 13864, loss: 8.775 I0415 23:26:32.258410 134049108950144 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2141.054, Tokens/s/device: 322722.975, total_weights: 12988, loss: 8.688 I0415 23:26:32.802303 133933964461632 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0`. I0415 23:26:32.867126 133933964461632 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 28.621616s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/0 I0415 23:26:32.867258 133933964461632 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 27.850153s. I0415 23:26:32.867417 133934046250560 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:26:32.867473 133934046250560 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:26:32.867535 133934046250560 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:26:32.867590 133934046250560 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0415 23:26:32.923676 134049108950144 train.py:632] Completed eval step 0 I0415 23:26:32.963160 134049108950144 train.py:632] Completed eval step 1 I0415 23:26:32.966691 134049108950144 metric_logger.py:204] eval metrics after step: 9, loss=8.653, total_weights=26703.0 I0415 23:26:32.967007 134049108950144 metric_logger.py:185] completed step: 8, seconds: 0.117, TFLOP/s/device: 115.912, Tokens/s/device: 17471.570, total_weights: 13820, loss: 8.740 I0415 23:26:32.969011 134049108950144 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0415 23:26:32.969606 134049108950144 checkpointing.py:776] Waited 0.0005970001220703125 seconds for step 9 to finish before starting checkpointing. I0415 23:26:32.969849 134049108950144 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:26:32.970020 134049108950144 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0415 23:26:32.970327 134049108950144 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9. I0415 23:26:33.162360 134049108950144 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array. I0415 23:26:33.162458 134049108950144 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 I0415 23:26:33.167306 133934046250560 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9 I0415 23:26:33.872367 133933243041344 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9/items I0415 23:26:42.316286 134049108950144 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.154348s I0415 23:26:42.320708 134049108950144 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.334 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host) I0415 23:26:42.320776 134049108950144 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.249956s (batch_requests_ready=0.088169s, total_serialization_initiated=9.157469s, others=0.004318s) I0415 23:26:42.320849 134049108950144 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.250499s (all_items=0.000026s, per_item={'items': '0.00002575'}, temp_paths=9.250473) I0415 23:26:42.322280 133934046250560 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0415 23:26:42.322361 134049108950144 async_checkpointer.py:561] Finished blocking save. Time taken: 9.352287s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9. I0415 23:26:42.322575 134049108950144 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0415 23:26:42.322773 133933964461632 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0415 23:26:42.322886 134049108950144 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776295592.96981, 'wait_for_prev_duration_secs': 6.4849853515625e-05, 'time_between_consecutive_saves_sec': 0.10220479965209961, 'checkpointer_blocking_start_time': 1776295592.9700413, 'checkpointer_blocking_duration_secs': 9.352424621582031, 'get_old_steps_start_time': 1776295602.3224905, 'get_old_steps_duration_secs': 4.220008850097656e-05, 'checkpoint_manager_blocking_start_time': 1776295592.9697673, 'checkpoint_manager_blocking_duration_secs': 9.353089809417725} I0415 23:26:42.323012 134049108950144 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0415 23:26:42.323070 134049108950144 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0415 23:26:43.067980 133933232555584 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9/items/array_metadatas/process_0 I0415 23:27:14.052055 133933243041344 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 31.730594s (commit=31.264934s, array_metadata_write=0.465661s) I0415 23:27:14.053119 133934046250560 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 308.348 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host) I0415 23:27:14.053174 133934046250560 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.730770s. I0415 23:27:14.508864 133934046250560 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0415 23:27:14.930398 133934046250560 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0415 23:27:14.930564 133934046250560 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.571439s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9/items I0415 23:27:14.931221 133934046250560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9/items I0415 23:27:15.192570 133934046250560 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9 I0415 23:27:15.880584 133934046250560 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9`. I0415 23:27:15.881231 133934046250560 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.911164s. directory=gs://wanglance-maxtext/linen_ckpt_feat_migrate_nnx_utils_20260415_230805/linen_feat_migrate_nnx_utils_20260415_230805_07_eval/checkpoints/9 I0415 23:27:15.881308 133934046250560 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.558905s. I0415 23:27:15.881500 133933964461632 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0415 23:27:15.881604 133933964461632 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0415 23:27:15.881677 133933964461632 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0415 23:27:15.881724 133933964461632 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0415 23:27:15.882112 134049108950144 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. I0415 23:27:15.882766 134049108950144 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0415 23:27:15.884194 134049108950144 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 115.727, Tokens/s/device: 17443.594, 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