2026-04-16 18:15:29.087182: 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 18:15:29.202937 130599023131776 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 18:16:19.587918 130599023131776 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 18:16:19.588024 130599023131776 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 18:16:19.588146 130599023131776 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 18:16:19.588174 130599023131776 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 18:16:19.588256 130599023131776 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 18:16:19.683537 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:16:20.090036 130599023131776 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 18:16:20.090221 130599023131776 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 18:16:20.090399 130599023131776 checkpointing.py:688] Setting up checkpoint logger...
I0416 18:16:20.090440 130599023131776 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 18:16:20.090476 130599023131776 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 18:16:20.091035 130599023131776 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 0x76c6dd7f5190>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 18:16:23.013894 130599023131776 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 18:16:23.014096 130599023131776 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 0x76c1319da9c0>}, handler_registry=None
I0416 18:16:23.014323 130599023131776 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76c1319da9c0>` 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 18:16:23.014361 130599023131776 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76c061556810>` 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 18:16:23.014389 130599023131776 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 0x76c1319da9c0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76c1319da9c0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76c061556810>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76c061556810>}).
I0416 18:16:23.014682 130599023131776 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 18:16:23.014735 130599023131776 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x76c06154f420> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 18:16:23.135913 130599023131776 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_01_base/checkpoints
I0416 18:16:23.136157 130599023131776 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_01_base/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x76c061556690>
I0416 18:16:23.136234 130599023131776 checkpointing.py:302] Checkpoint manager created!
I0416 18:16:23.216444 130599023131776 dataset_info.py:707] Load dataset info from tests/assets/local_datasets/c4_en_dataset_minimal/c4/en/3.1.0
I0416 18:16:23.219723 130599023131776 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 18:16:23.272539 130599023131776 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 18:16:23.298480 130599023131776 tokenizer.py:245] Tokenizer path: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 18:16:23.298544 130599023131776 tokenizer.py:187] Loading sentencepiece tokenizer: src/maxtext/assets/tokenizers/tokenizer.llama2
I0416 18:16:24.297506 130599023131776 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 18:16:24.297615 130599023131776 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:16:24.400834 130599023131776 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 18:16:24.400915 130599023131776 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:16:24.415912 130599023131776 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 18:16:24.415962 130599023131776 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:16:24.437483 130599023131776 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 18:16:24.437537 130599023131776 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:16:24.452808 130599023131776 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 18:16:24.452865 130599023131776 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:16:24.468128 130599023131776 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 18:16:24.468176 130599023131776 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:16:24.490405 130599023131776 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 18:16:24.490465 130599023131776 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:16:24.509373 130599023131776 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 18:16:24.509424 130599023131776 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 18:16:24.702690 130599023131776 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 18:16:24.702781 130599023131776 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0416 18:16:24.752841 3967547 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 18:16:25.147114 130599023131776 maxtext_utils.py:1796] params/params/decoder/decoder_norm/scale
Shape: float32[2048]
Logical: P('norm',)
Physical: (None,)
I0416 18:16:25.147220 130599023131776 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 18:16:25.147261 130599023131776 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 18:16:25.147305 130599023131776 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 18:16:25.147345 130599023131776 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 18:16:25.147372 130599023131776 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 18:16:25.147416 130599023131776 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 18:16:25.147454 130599023131776 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 18:16:25.147483 130599023131776 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 18:16:25.147511 130599023131776 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 18:16:25.147546 130599023131776 maxtext_utils.py:1796] params/params/decoder/logits_dense/kernel
Shape: float32[2048,32000]
Logical: P('embed_vocab', 'vocab')
Physical: ('fsdp', None)
I0416 18:16:25.147580 130599023131776 maxtext_utils.py:1796] params/params/token_embedder/embedding
Shape: float32[32000,2048]
Logical: P('vocab', 'embed_vocab')
Physical: (None, 'fsdp')
I0416 18:16:25.581881 130599023131776 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 18:16:25.581963 130599023131776 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 18:16:25.596627 130599023131776 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 18:16:25.596677 130599023131776 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 18:16:25.889319 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:16:26.003444 130599023131776 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 18:16:26.003916 130599023131776 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 18:16:26.004080 130599023131776 metric_logger.py:289] number parameters: 1.104 billion
W0416 18:16:29.077208 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:16:29.211197 130599023131776 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0416 18:16:29.331212 130599023131776 checkpointing.py:798] Waited 0.12000393867492676 seconds for step 0 to finish before starting checkpointing.
I0416 18:16:29.331629 130599023131776 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 18:16:29.331778 130599023131776 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 18:16:29.332150 130599023131776 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_01_base/checkpoints/0.
I0416 18:16:29.420933 130599023131776 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 18:16:29.516768 130599023131776 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 18:16:29.516900 130485781005888 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_01_base/checkpoints/0
I0416 18:16:29.517302 130599023131776 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 18:16:29.731720 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.745641 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.757968 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.775132 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.789717 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.796191 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.800656 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:29.804597 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:16:30.038228 130599023131776 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.522099s
I0416 18:16:30.038791 130599023131776 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 20.018 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 616 milliseconds) (per-host)
I0416 18:16:30.038851 130599023131776 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.616608s (batch_requests_ready=0.090040s, total_serialization_initiated=0.526168s, others=0.000401s)
I0416 18:16:30.038925 130599023131776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.617532s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=0.617512)
I0416 18:16:30.039930 130485709702720 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 18:16:30.040101 130599023131776 async_checkpointer.py:561] Finished blocking save. Time taken: 0.708276s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/0.
I0416 18:16:30.040278 130599023131776 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 18:16:30.040483 130485791491648 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 18:16:30.040599 130599023131776 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776363389.3316162, 'wait_for_prev_duration_secs': 4.172325134277344e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776363389.3317986, 'checkpointer_blocking_duration_secs': 0.708380937576294, 'get_old_steps_start_time': 1776363390.0402052, 'get_old_steps_duration_secs': 2.9802322387695312e-05, 'checkpoint_manager_blocking_start_time': 1776363389.331534, 'checkpoint_manager_blocking_duration_secs': 0.7090256214141846}
I0416 18:16:30.040701 130599023131776 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0416 18:16:30.040749 130599023131776 max_utils.py:750]
Memstats: After params initialized:
I0416 18:16:30.040859 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_0(process=0,(0,0,0,0))
I0416 18:16:30.040889 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 18:16:30.040909 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 18:16:30.040929 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 18:16:30.040948 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 18:16:30.040966 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 18:16:30.040990 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 18:16:30.041007 130599023131776 max_utils.py:756] Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 18:16:30.047293 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:30.052953 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:16:30.057713 3967547 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:16:30.366420 130599023131776 metric_logger.py:185] completed step: 0, seconds: 3.207, TFLOP/s/device: 4.237, Tokens/s/device: 638.683, total_weights: 13328, loss: 10.828
I0416 18:16:30.367308 130599023131776 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_01_base/tensorboard/'
I0416 18:16:30.440827 130485770520128 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_01_base/checkpoints/0/items
I0416 18:16:30.482211 130599023131776 metric_logger.py:185] completed step: 1, seconds: 1.153, TFLOP/s/device: 11.781, Tokens/s/device: 1775.732, total_weights: 12332, loss: 10.828
I0416 18:16:30.599689 130599023131776 metric_logger.py:185] completed step: 2, seconds: 0.009, TFLOP/s/device: 1472.221, Tokens/s/device: 221909.199, total_weights: 15161, loss: 9.823
I0416 18:16:30.601804 130485619525184 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363390354557676, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776363390354557676, "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_01_base/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:16:30.716822 130599023131776 metric_logger.py:185] completed step: 3, seconds: 0.116, TFLOP/s/device: 116.719, Tokens/s/device: 17593.141, total_weights: 13327, loss: 9.405
I0416 18:16:30.933610 3970809 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 18:16:30.933674 3970809 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 18:16:31.886979 130485730674240 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_01_base/checkpoints/0/items/array_metadatas/process_0
I0416 18:16:50.285773 130599023131776 metric_logger.py:185] completed step: 4, seconds: 0.118, TFLOP/s/device: 114.968, Tokens/s/device: 17329.204, total_weights: 11939, loss: 9.045
I0416 18:16:50.295246 130599023131776 metric_logger.py:185] completed step: 5, seconds: 0.117, TFLOP/s/device: 116.401, Tokens/s/device: 17545.212, total_weights: 15502, loss: 8.953
I0416 18:16:50.412901 130599023131776 metric_logger.py:185] completed step: 6, seconds: 19.569, TFLOP/s/device: 0.694, Tokens/s/device: 104.655, total_weights: 13864, loss: 8.775
I0416 18:16:50.530232 130599023131776 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2260.754, Tokens/s/device: 340765.391, total_weights: 12988, loss: 8.688
I0416 18:16:50.647586 130599023131776 metric_logger.py:185] completed step: 8, seconds: 0.119, TFLOP/s/device: 114.537, Tokens/s/device: 17264.198, total_weights: 13820, loss: 8.740
I0416 18:16:50.764960 130599023131776 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0416 18:16:50.765734 130599023131776 checkpointing.py:798] Waited 0.0008044242858886719 seconds for step 9 to finish before starting checkpointing.
I0416 18:16:50.766005 130599023131776 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 18:16:54.004281 130485720188480 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 23.066099s (commit=22.636811s, array_metadata_write=0.429289s)
I0416 18:16:54.005286 130485709702720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 514.046 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 24 seconds) (per-host)
I0416 18:16:54.005345 130485709702720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 23.965301s.
I0416 18:16:54.242561 130485709702720 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363390354557676, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:16:54.463482 130485709702720 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 18:16:54.631888 130485619525184 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776363390354557676, '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_01_base/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:16:54.848083 130485709702720 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 18:16:54.848576 130485709702720 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.552700s. 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_01_base/checkpoints/0/items
I0416 18:16:54.849220 130485709702720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/0/items
I0416 18:16:55.100678 130485709702720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/0
I0416 18:16:55.774937 130485709702720 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_01_base/checkpoints/0`.
I0416 18:16:55.775567 130485709702720 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 26.443756s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/0
I0416 18:16:55.775650 130485709702720 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 25.735606s.
I0416 18:16:55.775757 130485791491648 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 18:16:55.775806 130485791491648 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 18:16:55.775860 130485791491648 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 18:16:55.775901 130485791491648 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 18:16:55.776056 130599023131776 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 18:16:55.776184 130599023131776 checkpoint_manager.py:1441] Waiting for previous save to complete took 5.010192 seconds. If this number is high, consider checkpointing less frequently.
I0416 18:16:55.776978 130599023131776 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 18:16:55.777271 130599023131776 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_01_base/checkpoints/9.
I0416 18:16:55.953342 130599023131776 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 18:16:55.953482 130485791491648 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_01_base/checkpoints/9
I0416 18:16:55.953559 130599023131776 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 18:16:56.663245 130485720188480 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_01_base/checkpoints/9/items
I0416 18:17:04.132629 130599023131776 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 8.179985s
I0416 18:17:04.138054 130599023131776 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.491 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 8 seconds) (per-host)
I0416 18:17:04.138134 130599023131776 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 8.278855s (batch_requests_ready=0.090187s, total_serialization_initiated=8.183411s, others=0.005257s)
I0416 18:17:04.138227 130599023131776 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 8.279414s (all_items=0.000018s, per_item={'items': '0.00001836'}, temp_paths=8.279396)
I0416 18:17:04.140432 130485709702720 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 18:17:04.140578 130599023131776 async_checkpointer.py:561] Finished blocking save. Time taken: 8.363553s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/9.
I0416 18:17:04.140871 130599023131776 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 18:17:04.141090 130485567096384 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 18:17:04.141237 130599023131776 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_01_base/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776363410.765961, 'wait_for_prev_duration_secs': 5.010192394256592, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776363415.7770007, 'checkpointer_blocking_duration_secs': 8.36373233795166, 'get_old_steps_start_time': 1776363424.140756, 'get_old_steps_duration_secs': 5.984306335449219e-05, 'checkpoint_manager_blocking_start_time': 1776363410.7659202, 'checkpoint_manager_blocking_duration_secs': 13.375279188156128}
I0416 18:17:04.141390 130599023131776 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0416 18:17:04.141431 130599023131776 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 18:17:04.898059 130485730674240 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_01_base/checkpoints/9/items/array_metadatas/process_0
I0416 18:17:42.920958 130485720188480 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.781862s (commit=38.368958s, array_metadata_write=0.412903s)
I0416 18:17:42.922621 130485709702720 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 268.507 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 47 seconds) (per-host)
I0416 18:17:42.922800 130485709702720 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.782152s.
I0416 18:17:43.382999 130485709702720 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 18:17:43.786164 130485709702720 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 18:17:43.786413 130485709702720 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.560766s. 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_01_base/checkpoints/9/items
I0416 18:17:43.788156 130485709702720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/9/items
I0416 18:17:44.044853 130485709702720 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/9
I0416 18:17:44.739184 130485709702720 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_01_base/checkpoints/9`.
I0416 18:17:44.740103 130485709702720 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 48.963084s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_01_base/checkpoints/9
I0416 18:17:44.740203 130485709702720 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 40.599565s.
I0416 18:17:44.740534 130485567096384 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 18:17:44.740708 130485567096384 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 18:17:44.740820 130485567096384 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 18:17:44.740888 130485567096384 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 18:17:44.741236 130599023131776 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 18:17:44.741466 130599023131776 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 18:17:44.742318 130599023131776 metric_logger.py:185] completed step: 9, seconds: 0.117, TFLOP/s/device: 115.772, Tokens/s/device: 17450.431, 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