MaxView

2026-04-16 20:18:59.099860: 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 20:18:59.214424 137240024468608 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 20:19:55.445519 137240024468608 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 20:19:55.445628 137240024468608 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 20:19:55.445662 137240024468608 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 20:19:55.445687 137240024468608 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 20:19:55.445708 137240024468608 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 20:19:55.445780 137240024468608 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 20:19:55.538401 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:19:55.926523 137240024468608 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 20:19:56.024760 137240024468608 checkpointing.py:688] Setting up checkpoint logger...
I0416 20:19:56.024857 137240024468608 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 20:19:56.024916 137240024468608 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 20:19:56.025159 137240024468608 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 0x7cd1189f8ec0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 20:19:58.457904 137240024468608 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 20:19:58.458134 137240024468608 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 0x7cca98911850>}, handler_registry=None
I0416 20:19:58.458458 137240024468608 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cca98911850>` 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 20:19:58.458498 137240024468608 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cca989844a0>` 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 20:19:58.458527 137240024468608 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 0x7cca98911850>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7cca98911850>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cca989844a0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7cca989844a0>}).
I0416 20:19:58.458843 137240024468608 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 20:19:58.458907 137240024468608 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7cca99975760> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 20:19:58.590288 137240024468608 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints
I0416 20:19:58.590546 137240024468608 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/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7cca98913d10>
I0416 20:19:58.590644 137240024468608 checkpointing.py:302] Checkpoint manager created!
W0416 20:19:58.606776 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:19:58.814823 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='count'), GetAttrKey(name='value')) to host
I0416 20:19:58.814921 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.814954 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.814980 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815001 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815023 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.815056 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.815080 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815103 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815121 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815138 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815155 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815172 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to host
I0416 20:19:58.815192 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.815210 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815227 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815243 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815260 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.815277 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to host
I0416 20:19:58.815294 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815313 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815330 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815347 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815363 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to host
I0416 20:19:58.815380 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to host
I0416 20:19:58.815396 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=2), DictKey(key='count'), GetAttrKey(name='value')) to host
I0416 20:19:58.815418 137240024468608 maxtext_utils_nnx.py:112] max_utils.py: Moving (DictKey(key='step'), GetAttrKey(name='value')) to host
I0416 20:19:58.816293 137240024468608 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 20:19:58.816349 137240024468608 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0416 20:19:58.927487 4095147 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 20:19:59.035853 137240024468608 maxtext_utils.py:1805]  decoder/decoder_norm/scale/value
    Shape:     float32[2048]
    Physical:  (None,)
I0416 20:19:59.035961 137240024468608 maxtext_utils.py:1805]  decoder/layers/mlp/wi_0/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:19:59.036003 137240024468608 maxtext_utils.py:1805]  decoder/layers/mlp/wi_1/kernel/value
    Shape:     float32[2048,16,7168]
    Physical:  ('fsdp', None, None)
I0416 20:19:59.036035 137240024468608 maxtext_utils.py:1805]  decoder/layers/mlp/wo/kernel/value
    Shape:     float32[7168,16,2048]
    Physical:  (None, None, 'fsdp')
I0416 20:19:59.036080 137240024468608 maxtext_utils.py:1805]  decoder/layers/post_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:19:59.036107 137240024468608 maxtext_utils.py:1805]  decoder/layers/pre_self_attention_layer_norm/scale/value
    Shape:     float32[2048,16]
    Physical:  (None, None)
I0416 20:19:59.036132 137240024468608 maxtext_utils.py:1805]  decoder/layers/self_attention/key/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:19:59.036159 137240024468608 maxtext_utils.py:1805]  decoder/layers/self_attention/out/kernel/value
    Shape:     float32[16,16,128,2048]
    Physical:  (None, None, None, 'fsdp')
I0416 20:19:59.036184 137240024468608 maxtext_utils.py:1805]  decoder/layers/self_attention/query/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:19:59.036207 137240024468608 maxtext_utils.py:1805]  decoder/layers/self_attention/value/kernel/value
    Shape:     float32[2048,16,16,128]
    Physical:  ('fsdp', None, None, None)
I0416 20:19:59.036229 137240024468608 maxtext_utils.py:1805]  decoder/logits_dense/kernel/value
    Shape:     float32[2048,32000]
    Physical:  ('fsdp', None)
I0416 20:19:59.036252 137240024468608 maxtext_utils.py:1805]  token_embedder/embedding/value
    Shape:     float32[32000,2048]
    Physical:  (None, 'fsdp')

I0416 20:19:59.144969 137240024468608 nnx_decoders.py:465] nnx_decoders/carry Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:19:59.145059 137240024468608 nnx_decoders.py:465] nnx_decoders/carry Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:19:59.163090 137240024468608 nnx_decoders.py:465] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 20:19:59.163149 137240024468608 nnx_decoders.py:465] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:19:59.178230 137240024468608 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:19:59.178280 137240024468608 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:19:59.192025 137240024468608 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 20:19:59.192086 137240024468608 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 20:19:59.242957 137240024468608 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 20:19:59.243031 137240024468608 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:19:59.257034 137240024468608 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 20:19:59.257098 137240024468608 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:19:59.275493 137240024468608 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 20:19:59.275576 137240024468608 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:19:59.381102 137240024468608 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 20:19:59.381182 137240024468608 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 20:19:59.404733 137240024468608 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 20:19:59.404791 137240024468608 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 20:19:59.877543 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='count'), GetAttrKey(name='value')) to device
I0416 20:19:59.877653 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.877688 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877720 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877816 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877844 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.877866 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.877890 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877911 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877931 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877948 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877966 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.877986 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to device
I0416 20:19:59.878009 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.878029 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878064 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878086 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878103 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.878120 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:19:59.878139 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878157 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878175 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878192 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878209 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:19:59.878226 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to device
I0416 20:19:59.878247 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=2), DictKey(key='count'), GetAttrKey(name='value')) to device
I0416 20:19:59.878266 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='step'), GetAttrKey(name='value')) to device
W0416 20:20:00.563017 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:20:00.684374 137240024468608 max_utils.py:791] Total memory size: 2.4 GB, Output size: 0.5 GB, Temp size: 1.9 GB, Argument size: 0.5 GB, Host temp size: 0.0 GB.
I0416 20:20:03.462700 137240024468608 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 20:20:03.464387 137240024468608 metric_logger.py:289] number parameters: 1.104 billion
I0416 20:20:03.878883 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='count'), GetAttrKey(name='value')) to device
I0416 20:20:03.878979 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879013 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879039 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879158 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879187 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879210 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879229 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879249 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879269 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879289 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879307 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879326 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='mu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to device
I0416 20:20:03.879346 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='decoder_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879369 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_0'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879388 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wi_1'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879406 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='mlp'), DictKey(key='wo'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879424 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='post_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879442 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='pre_self_attention_layer_norm'), DictKey(key='scale'), GetAttrKey(name='value')) to device
I0416 20:20:03.879460 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='key'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879477 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='out'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879494 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='query'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879511 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='layers'), DictKey(key='self_attention'), DictKey(key='value'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879530 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='decoder'), DictKey(key='logits_dense'), DictKey(key='kernel'), GetAttrKey(name='value')) to device
I0416 20:20:03.879548 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=0), DictKey(key='nu'), DictKey(key='token_embedder'), DictKey(key='embedding'), GetAttrKey(name='value')) to device
I0416 20:20:03.879565 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='opt_state'), DictKey(key=2), DictKey(key='count'), GetAttrKey(name='value')) to device
I0416 20:20:03.879582 137240024468608 maxtext_utils_nnx.py:129] max_utils.py: Moving (DictKey(key='step'), GetAttrKey(name='value')) to device
W0416 20:20:04.071518 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:20:04.193138 137240024468608 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0416 20:20:04.377686 137240024468608 checkpointing.py:798] Waited 0.18454337120056152 seconds for step 0 to finish before starting checkpointing.
I0416 20:20:04.378169 137240024468608 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:20:04.378339 137240024468608 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 20:20:04.378758 137240024468608 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0.
I0416 20:20:04.464292 137240024468608 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 20:20:04.563321 137132677006912 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0
I0416 20:20:04.563865 137240024468608 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:20:04.564105 137240024468608 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 20:20:05.196306 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.205067 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.215380 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.227942 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.234974 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.242537 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.249520 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.253502 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:20:05.380423 137132666521152 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/items
I0416 20:20:05.428770 137240024468608 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.866502s
I0416 20:20:05.429110 137240024468608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.811 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 963 milliseconds) (per-host)
I0416 20:20:05.429163 137240024468608 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.963387s (batch_requests_ready=0.087845s, total_serialization_initiated=0.875295s, others=0.000247s)
I0416 20:20:05.429225 137240024468608 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.964474s (all_items=0.000023s, per_item={'items': '0.00002265'}, temp_paths=0.964451)
I0416 20:20:05.429958 137132605703744 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:20:05.430066 137240024468608 async_checkpointer.py:561] Finished blocking save. Time taken: 1.051677s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0.
I0416 20:20:05.430258 137240024468608 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:20:05.430437 137240024468608 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776370804.3781528, 'wait_for_prev_duration_secs': 4.4345855712890625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776370804.3783605, 'checkpointer_blocking_duration_secs': 1.0518181324005127, 'get_old_steps_start_time': 1776370805.4301987, 'get_old_steps_duration_secs': 2.1457672119140625e-05, 'checkpoint_manager_blocking_start_time': 1776370804.3780358, 'checkpoint_manager_blocking_duration_secs': 1.0523762702941895}
I0416 20:20:05.430506 137240024468608 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0416 20:20:05.430556 137240024468608 max_utils.py:750] 
Memstats: After params initialized:
I0416 20:20:05.430672 137240024468608 max_utils.py:756] 	Using (GB) 0.57 / 31.25 (1.824000%) on TPU_0(process=0,(0,0,0,0))
I0416 20:20:05.430708 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_1(process=0,(1,0,0,0))
I0416 20:20:05.430730 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_2(process=0,(0,1,0,0))
I0416 20:20:05.430749 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_3(process=0,(1,1,0,0))
I0416 20:20:05.430768 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_4(process=0,(0,2,0,0))
I0416 20:20:05.430787 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_5(process=0,(1,2,0,0))
I0416 20:20:05.430803 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_6(process=0,(0,3,0,0))
I0416 20:20:05.430819 137240024468608 max_utils.py:756] 	Using (GB) 0.56 / 31.25 (1.792000%) on TPU_7(process=0,(1,3,0,0))
W0416 20:20:05.436577 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.442180 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 20:20:05.446812 4095147 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 20:20:05.515349 137132626675264 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370805241270014, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776370805241270014, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 20:20:05.515580 137132687492672 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:20:05.768459 137240024468608 metric_logger.py:185] completed step: 0, seconds: 0.727, TFLOP/s/device: 18.677, Tokens/s/device: 2815.252, total_weights: 16384, loss: 10.881
I0416 20:20:05.769219 137240024468608 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/tensorboard/'
I0416 20:20:05.949600 137240024468608 metric_logger.py:185] completed step: 1, seconds: 1.571, TFLOP/s/device: 8.647, Tokens/s/device: 1303.319, total_weights: 16384, loss: 10.881
I0416 20:20:06.096073 4098057 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 20:20:06.096148 4098057 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 20:20:06.137320 137240024468608 metric_logger.py:185] completed step: 2, seconds: 0.019, TFLOP/s/device: 709.807, Tokens/s/device: 106989.865, total_weights: 16384, loss: 9.779
I0416 20:20:06.324692 137240024468608 metric_logger.py:185] completed step: 3, seconds: 0.173, TFLOP/s/device: 78.642, Tokens/s/device: 11853.841, total_weights: 16384, loss: 8.837
I0416 20:20:06.912234 137132677006912 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_0
I0416 20:20:26.562910 137240024468608 metric_logger.py:185] completed step: 4, seconds: 0.188, TFLOP/s/device: 72.214, Tokens/s/device: 10884.817, total_weights: 16384, loss: 7.961
I0416 20:20:26.572877 137240024468608 metric_logger.py:185] completed step: 5, seconds: 0.187, TFLOP/s/device: 72.705, Tokens/s/device: 10958.904, total_weights: 16384, loss: 7.226
I0416 20:20:26.756037 137240024468608 metric_logger.py:185] completed step: 6, seconds: 20.240, TFLOP/s/device: 0.671, Tokens/s/device: 101.187, total_weights: 16384, loss: 6.669
I0416 20:20:26.942978 137240024468608 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 2148.842, Tokens/s/device: 323896.884, total_weights: 16384, loss: 6.286
I0416 20:20:27.129832 137240024468608 metric_logger.py:185] completed step: 8, seconds: 0.185, TFLOP/s/device: 73.598, Tokens/s/device: 11093.477, total_weights: 16384, loss: 6.048
I0416 20:20:27.317003 137240024468608 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0416 20:20:27.318205 137240024468608 checkpointing.py:798] Waited 0.0012133121490478516 seconds for step 9 to finish before starting checkpointing.
I0416 20:20:27.318468 137240024468608 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:20:34.779875 137132616189504 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 28.695023s (commit=28.191582s, array_metadata_write=0.503441s)
I0416 20:20:34.781097 137132605703744 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 416.846 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 30 seconds) (per-host)
I0416 20:20:34.781205 137132605703744 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 29.351094s.
I0416 20:20:35.007835 137132605703744 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370805241270014, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 20:20:35.172494 137132605703744 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 20:20:35.418149 137132626675264 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776370805241270014, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/_CHECKPOINT_METADATA
I0416 20:20:35.622770 137132605703744 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 20:20:35.623370 137132605703744 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.575045s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/items
I0416 20:20:35.624058 137132605703744 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0/items
I0416 20:20:35.876140 137132605703744 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0
I0416 20:20:36.595066 137132605703744 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0`.
I0416 20:20:36.595681 137132605703744 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 32.217309s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/0
I0416 20:20:36.595757 137132605703744 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 31.165648s.
I0416 20:20:36.595943 137132687492672 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:20:36.596060 137132687492672 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:20:36.596121 137132687492672 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:20:36.596166 137132687492672 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 20:20:36.596308 137240024468608 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 20:20:36.596427 137240024468608 checkpoint_manager.py:1441] Waiting for previous save to complete took 9.277976 seconds. If this number is high, consider checkpointing less frequently.
I0416 20:20:36.597205 137240024468608 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 20:20:36.597487 137240024468608 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9.
I0416 20:20:36.775807 137132687492672 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9
I0416 20:20:36.817739 137240024468608 jax_array_handlers.py:347] Scheduling D2H of 69 prioritized jax.Array.
I0416 20:20:36.817890 137240024468608 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 20:20:37.029537 137240024468608 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.212795s
I0416 20:20:37.029951 137240024468608 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 35.969 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 343 milliseconds) (per-host)
I0416 20:20:37.030001 137240024468608 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.343176s (batch_requests_ready=0.122970s, total_serialization_initiated=0.219885s, others=0.000321s)
I0416 20:20:37.030080 137240024468608 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.343908s (all_items=0.000018s, per_item={'items': '0.00001812'}, temp_paths=0.343890)
I0416 20:20:37.031010 137132565857856 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 20:20:37.031110 137240024468608 async_checkpointer.py:561] Finished blocking save. Time taken: 0.433861s. Continuing background save to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9.
I0416 20:20:37.031280 137240024468608 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 20:20:37.031403 137132605703744 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 20:20:37.031491 137240024468608 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776370827.318425, 'wait_for_prev_duration_secs': 9.277976274490356, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776370836.5972316, 'checkpointer_blocking_duration_secs': 0.43397021293640137, 'get_old_steps_start_time': 1776370837.0312157, 'get_old_steps_duration_secs': 2.7179718017578125e-05, 'checkpoint_manager_blocking_start_time': 1776370827.3183863, 'checkpoint_manager_blocking_duration_secs': 9.713074445724487}
I0416 20:20:37.031573 137240024468608 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0416 20:20:37.031604 137240024468608 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 20:20:37.534607 137132494554688 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9/items
I0416 20:20:38.856706 137132586829376 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 69 array_metadata.ArrayMetadata to gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_0
I0416 20:21:16.911210 137132576343616 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 38.897429s (commit=38.436058s, array_metadata_write=0.461371s)
I0416 20:21:16.912287 137132565857856 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 314.149 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 40 seconds) (per-host)
I0416 20:21:16.912396 137132565857856 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 39.881235s.
I0416 20:21:17.376929 137132565857856 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 20:21:17.804076 137132565857856 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 20:21:17.804261 137132565857856 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.567528s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9/items
I0416 20:21:17.804891 137132565857856 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9/items
I0416 20:21:18.040484 137132565857856 atomicity.py:608] Finalizing gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9
I0416 20:21:18.717416 137132565857856 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9`.
I0416 20:21:18.718064 137132565857856 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 42.120799s. directory=gs://wanglance-maxtext/nnx_ckpt_feat_nnx_post_train_fixes_20260416_181521/nnx_feat_nnx_post_train_fixes_20260416_181521_11_optimizer_offload_true/checkpoints/9
I0416 20:21:18.718142 137132565857856 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 41.686987s.
I0416 20:21:18.718261 137132605703744 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 20:21:18.718313 137132605703744 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 20:21:18.718387 137132605703744 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 20:21:18.718430 137132605703744 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 20:21:18.718698 137240024468608 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 20:21:18.719263 137240024468608 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 20:21:18.719932 137240024468608 metric_logger.py:185] completed step: 9, seconds: 0.186, TFLOP/s/device: 72.888, Tokens/s/device: 10986.417, total_weights: 16384, loss: 5.908
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops