MaxView

2026-04-16 18:29:32.350604: 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 18:29:32.465714 130576487066752 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag.
I0416 18:30:27.309549 130576487066752 max_utils.py:800] System Information: Jax Version: 0.9.2
I0416 18:30:27.309664 130576487066752 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0416 18:30:27.309697 130576487066752 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 18:30:27.309722 130576487066752 train_utils.py:364] WARNING: 'dataset_path' might be pointing your local file system
I0416 18:30:27.309743 130576487066752 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0416 18:30:27.309813 130576487066752 train.py:811] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper.
W0416 18:30:27.401406 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:30:27.866101 130576487066752 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 18:30:27.866284 130576487066752 maxtext_utils.py:1687] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0416 18:30:27.866529 130576487066752 checkpointing.py:688] Setting up checkpoint logger...
I0416 18:30:27.866581 130576487066752 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0416 18:30:27.866622 130576487066752 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None
I0416 18:30:27.867330 130576487066752 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 0x76c19e3f4e60>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0416 18:30:30.215345 130576487066752 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0416 18:30:30.215518 130576487066752 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 0x76bb1e935fa0>}, handler_registry=None
I0416 18:30:30.215772 130576487066752 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76bb1e935fa0>` 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 18:30:30.215812 130576487066752 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76bb1e9365d0>` 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 18:30:30.215840 130576487066752 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 0x76bb1e935fa0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x76bb1e935fa0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76bb1e9365d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x76bb1e9365d0>}).
I0416 18:30:30.216142 130576487066752 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28
I0416 18:30:30.216192 130576487066752 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x76bb1e931ee0> timeout: 600 secs and primary_host=0 for async checkpoint writes
I0416 18:30:30.333783 130576487066752 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_06_grad_accum/checkpoints
I0416 18:30:30.333997 130576487066752 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_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x76bb1e934ef0>
I0416 18:30:30.334089 130576487066752 checkpointing.py:302] Checkpoint manager created!
I0416 18:30:31.234030 130576487066752 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0416 18:30:31.234206 130576487066752 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:30:31.324098 130576487066752 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 18:30:31.324179 130576487066752 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:30:31.338730 130576487066752 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0416 18:30:31.338781 130576487066752 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None).
I0416 18:30:31.365243 130576487066752 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 18:30:31.365329 130576487066752 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:30:31.379916 130576487066752 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 18:30:31.379973 130576487066752 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:30:31.394345 130576487066752 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 18:30:31.394391 130576487066752 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:30:31.416425 130576487066752 attentions.py:1197] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0416 18:30:31.416482 130576487066752 attentions.py:1197] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None).
I0416 18:30:31.434627 130576487066752 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0416 18:30:31.434679 130576487066752 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None).
I0416 18:30:31.625440 130576487066752 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0416 18:30:31.625530 130576487066752 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
W0416 18:30:31.676442 3984498 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 18:30:32.061659 130576487066752 maxtext_utils.py:1796]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0416 18:30:32.061770 130576487066752 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 18:30:32.061810 130576487066752 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 18:30:32.061853 130576487066752 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 18:30:32.061903 130576487066752 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 18:30:32.061932 130576487066752 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 18:30:32.061972 130576487066752 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 18:30:32.062010 130576487066752 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 18:30:32.062038 130576487066752 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 18:30:32.062081 130576487066752 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 18:30:32.062118 130576487066752 maxtext_utils.py:1796]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0416 18:30:32.062151 130576487066752 maxtext_utils.py:1796]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0416 18:30:32.081286 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown.
I0416 18:30:32.081345 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,).
I0416 18:30:32.094566 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown.
I0416 18:30:32.094617 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None).
I0416 18:30:32.120657 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown.
I0416 18:30:32.120714 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp').
I0416 18:30:32.133598 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown.
I0416 18:30:32.133644 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None).
I0416 18:30:32.159559 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown.
I0416 18:30:32.159612 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None).
I0416 18:30:32.172587 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown.
I0416 18:30:32.172636 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp').
I0416 18:30:32.211696 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown.
I0416 18:30:32.211759 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None).
I0416 18:30:32.224758 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown.
I0416 18:30:32.224803 130576487066752 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp').
I0416 18:30:32.805700 130576487066752 train.py:157] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 18:30:32.805783 130576487066752 train.py:157] train/xent Physical: float32[8,2048]............................................. ('fsdp', None).
I0416 18:30:32.819559 130576487066752 train.py:164] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length').
I0416 18:30:32.819606 130576487066752 train.py:164] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None).
W0416 18:30:33.637385 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:30:33.778435 130576487066752 max_utils.py:791] Total memory size: 4.5 GB, Output size: 1.5 GB, Temp size: 2.9 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB.
I0416 18:30:33.778922 130576487066752 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter.
I0416 18:30:33.779088 130576487066752 metric_logger.py:289] number parameters: 1.104 billion
W0416 18:30:35.146114 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:30:35.281485 130576487066752 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0416 18:30:35.693120 130576487066752 checkpointing.py:798] Waited 0.4116685390472412 seconds for step 0 to finish before starting checkpointing.
I0416 18:30:35.693606 130576487066752 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 18:30:35.693756 130576487066752 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0
I0416 18:30:35.694226 130576487066752 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_06_grad_accum/checkpoints/0.
I0416 18:30:35.776018 130576487066752 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient
I0416 18:30:35.868257 130576487066752 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 18:30:35.868415 130468836017728 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_06_grad_accum/checkpoints/0
I0416 18:30:35.868792 130576487066752 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 18:30:36.104951 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.118194 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.132696 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.148770 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.157487 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.163230 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.168288 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.172669 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:30:36.422972 130576487066752 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.555434s
I0416 18:30:36.423332 130576487066752 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 19.103 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 645 milliseconds) (per-host)
I0416 18:30:36.423386 130576487066752 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.646094s (batch_requests_ready=0.086851s, total_serialization_initiated=0.558988s, others=0.000254s)
I0416 18:30:36.423444 130576487066752 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.646957s (all_items=0.000022s, per_item={'items': '0.00002217'}, temp_paths=0.646935)
I0416 18:30:36.424687 130468773103168 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 18:30:36.424782 130576487066752 async_checkpointer.py:561] Finished blocking save. Time taken: 0.730986s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/0.
I0416 18:30:36.424950 130576487066752 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 18:30:36.425099 130468846503488 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 18:30:36.425171 130576487066752 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_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776364235.693584, 'wait_for_prev_duration_secs': 5.030632019042969e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776364235.6937773, 'checkpointer_blocking_duration_secs': 0.7311050891876221, 'get_old_steps_start_time': 1776364236.4248989, 'get_old_steps_duration_secs': 2.002716064453125e-05, 'checkpoint_manager_blocking_start_time': 1776364235.6934981, 'checkpoint_manager_blocking_duration_secs': 0.7316389083862305}
I0416 18:30:36.425279 130576487066752 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0416 18:30:36.425327 130576487066752 max_utils.py:750] 
Memstats: After params initialized:
I0416 18:30:36.425410 130576487066752 max_utils.py:756] 	Using (GB) 1.6 / 31.25 (5.120000%) on TPU_0(process=0,(0,0,0,0))
I0416 18:30:36.425437 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_1(process=0,(1,0,0,0))
I0416 18:30:36.425456 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_2(process=0,(0,1,0,0))
I0416 18:30:36.425473 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_3(process=0,(1,1,0,0))
I0416 18:30:36.425491 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_4(process=0,(0,2,0,0))
I0416 18:30:36.425508 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_5(process=0,(1,2,0,0))
I0416 18:30:36.425524 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_6(process=0,(0,3,0,0))
I0416 18:30:36.425540 130576487066752 max_utils.py:756] 	Using (GB) 1.59 / 31.25 (5.088000%) on TPU_7(process=0,(1,3,0,0))
W0416 18:30:36.431248 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.437083 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
W0416 18:30:36.441830 3984498 pjrt_executable.cc:642] Assume version compatibility. PjRt-IFRT does not track XLA executable versions.
I0416 18:30:36.577798 130468825531968 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_06_grad_accum/checkpoints/0/items
I0416 18:30:36.753598 130468754228800 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776364236492831810, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776364236492831810, "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_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:30:36.760639 130576487066752 metric_logger.py:185] completed step: 0, seconds: 1.502, TFLOP/s/device: 36.187, Tokens/s/device: 5454.541, total_weights: 65536, loss: 10.871
I0416 18:30:36.761638 130576487066752 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_06_grad_accum/tensorboard/'
I0416 18:30:37.143735 3986746 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json
I0416 18:30:37.143798 3986746 google_auth_provider.cc:156] Using OAuth2 AuthProvider
I0416 18:30:37.186939 130576487066752 metric_logger.py:185] completed step: 1, seconds: 1.477, TFLOP/s/device: 36.790, Tokens/s/device: 5545.368, total_weights: 65536, loss: 10.871
I0416 18:30:37.619924 130576487066752 metric_logger.py:185] completed step: 2, seconds: 0.007, TFLOP/s/device: 7931.774, Tokens/s/device: 1195563.339, total_weights: 65536, loss: 10.261
I0416 18:30:37.885959 130468794074688 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_06_grad_accum/checkpoints/0/items/array_metadatas/process_0
I0416 18:30:38.049309 130576487066752 metric_logger.py:185] completed step: 3, seconds: 0.427, TFLOP/s/device: 127.187, Tokens/s/device: 19170.959, total_weights: 65536, loss: 9.732
I0416 18:31:07.754033 130468783588928 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 30.615618s (commit=30.193714s, array_metadata_write=0.421904s)
I0416 18:31:07.755132 130468773103168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 395.174 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 31 seconds) (per-host)
I0416 18:31:07.755193 130468773103168 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 31.330363s.
I0416 18:31:08.018337 130468773103168 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776364236492831810, '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_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:31:08.226293 130468773103168 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 18:31:08.430444 130468754228800 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776364236492831810, '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_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0416 18:31:08.632231 130468773103168 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 18:31:08.633078 130468773103168 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.551467s. 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_06_grad_accum/checkpoints/0/items
I0416 18:31:08.633733 130468773103168 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/0/items
I0416 18:31:08.860464 130468773103168 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/0
I0416 18:31:09.500504 130468773103168 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_06_grad_accum/checkpoints/0`.
I0416 18:31:09.501169 130468773103168 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 33.807378s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/0
I0416 18:31:09.501250 130468773103168 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 33.076424s.
I0416 18:31:09.501383 130468846503488 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 18:31:09.501440 130468846503488 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 18:31:09.501524 130468846503488 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0416 18:31:09.501581 130468846503488 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0416 18:31:14.548251 130576487066752 metric_logger.py:185] completed step: 4, seconds: 0.434, TFLOP/s/device: 125.295, Tokens/s/device: 18885.759, total_weights: 65536, loss: 9.274
I0416 18:31:14.555147 130576487066752 metric_logger.py:185] completed step: 5, seconds: 0.429, TFLOP/s/device: 126.744, Tokens/s/device: 19104.210, total_weights: 65536, loss: 8.888
I0416 18:31:14.977724 130576487066752 metric_logger.py:185] completed step: 6, seconds: 36.499, TFLOP/s/device: 1.489, Tokens/s/device: 224.447, total_weights: 65536, loss: 8.588
I0416 18:31:15.391730 130576487066752 metric_logger.py:185] completed step: 7, seconds: 0.005, TFLOP/s/device: 11084.747, Tokens/s/device: 1670813.787, total_weights: 65536, loss: 8.380
I0416 18:31:15.825266 130576487066752 metric_logger.py:185] completed step: 8, seconds: 0.423, TFLOP/s/device: 128.461, Tokens/s/device: 19362.997, total_weights: 65536, loss: 8.251
I0416 18:31:16.252530 130576487066752 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0416 18:31:16.253324 130576487066752 checkpointing.py:798] Waited 0.0008215904235839844 seconds for step 9 to finish before starting checkpointing.
I0416 18:31:16.253599 130576487066752 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 18:31:16.253772 130576487066752 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9
I0416 18:31:16.254088 130576487066752 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_06_grad_accum/checkpoints/9.
I0416 18:31:16.415682 130468846503488 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_06_grad_accum/checkpoints/9
I0416 18:31:16.432453 130576487066752 jax_array_handlers.py:347] Scheduling D2H of 39 prioritized jax.Array.
I0416 18:31:16.432547 130576487066752 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 18:31:17.129304 130468825531968 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_06_grad_accum/checkpoints/9/items
I0416 18:31:26.300699 130576487066752 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 9.868641s
I0416 18:31:26.306607 130576487066752 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.238 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 9 seconds) (per-host)
I0416 18:31:26.306700 130576487066752 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 9.968348s (batch_requests_ready=0.090755s, total_serialization_initiated=9.871930s, others=0.005662s)
I0416 18:31:26.306799 130576487066752 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 9.968924s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=9.968904)
I0416 18:31:26.309079 130468773103168 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started.
I0416 18:31:26.309228 130576487066752 async_checkpointer.py:561] Finished blocking save. Time taken: 10.055408s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/9.
I0416 18:31:26.309578 130576487066752 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0416 18:31:26.309803 130468846503488 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0416 18:31:26.309974 130576487066752 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_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776364276.2535508, 'wait_for_prev_duration_secs': 7.295608520507812e-05, 'time_between_consecutive_saves_sec': 6.751953840255737, 'checkpointer_blocking_start_time': 1776364276.2537942, 'checkpointer_blocking_duration_secs': 10.055597066879272, 'get_old_steps_start_time': 1776364286.3094282, 'get_old_steps_duration_secs': 8.797645568847656e-05, 'checkpoint_manager_blocking_start_time': 1776364276.2535074, 'checkpoint_manager_blocking_duration_secs': 10.056419849395752}
I0416 18:31:26.310154 130576487066752 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0416 18:31:26.310209 130576487066752 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0416 18:31:27.028480 130468836017728 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_06_grad_accum/checkpoints/9/items/array_metadatas/process_0
I0416 18:32:01.606280 130468825531968 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 35.298563s (commit=34.860293s, array_metadata_write=0.438269s)
I0416 18:32:01.608080 130468773103168 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 279.145 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 45 seconds) (per-host)
I0416 18:32:01.608264 130468773103168 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 35.298970s.
I0416 18:32:02.034524 130468773103168 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 18:32:02.451422 130468773103168 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389).
I0416 18:32:02.451668 130468773103168 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.567713s. 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_06_grad_accum/checkpoints/9/items
I0416 18:32:02.452551 130468773103168 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/9/items
I0416 18:32:02.697060 130468773103168 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/9
I0416 18:32:03.364960 130468773103168 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_06_grad_accum/checkpoints/9`.
I0416 18:32:03.365830 130468773103168 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 47.112023s. directory=gs://wanglance-maxtext/linen_ckpt_feat_nnx_post_train_fixes_20260416_181521/linen_feat_nnx_post_train_fixes_20260416_181521_06_grad_accum/checkpoints/9
I0416 18:32:03.365927 130468773103168 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 37.056638s.
I0416 18:32:03.366228 130468846503488 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0416 18:32:03.366398 130468846503488 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0416 18:32:03.366536 130468846503488 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0416 18:32:03.366593 130468846503488 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0416 18:32:03.367037 130576487066752 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 18:32:03.367211 130576487066752 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0416 18:32:03.368160 130576487066752 metric_logger.py:185] completed step: 9, seconds: 0.414, TFLOP/s/device: 131.277, Tokens/s/device: 19787.440, total_weights: 65536, loss: 8.175
Per train step:
 Total TFLOPs: 54.35 
 split as 93.93% learnable weight flops and 6.07% attention flops