MaxView

‹ 11_optimizer_offload_trueCase: 13_scan_layers_false13_scan_layers_true ›

Metrics: main (7f7822822) vs feat/nnx-post-train-fixes (2fbc8c2f1)

Metricmain  7f7822822feat/nnx-post-train-fixes  2fbc8c2f1Diff (feat/nnx-post-train-fixes − main)
Parameters1.104 billion1.104 billion
Final loss8.18808.18800
TFLOP/s97.790100.462+2.672
Tok/s14740.015142.8+402.814
Avg s/step3.0532.939-0.114
Memory %2.622.620
JAX0.9.20.9.2

Diff = branch value − main value. Green = branch improved. Red = branch regressed.

main  ·  7f7822822  ·  main_20260417_125726  ·  full log
XPK Start: Fri Apr 17 14:09:12 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 14:09:36.921730: 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)
I0417 14:09:37.100418 133882125645632 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 14:09:46.140721 133882125645632 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 14:09:46.143069 133882125645632 distributed.py:172] Connecting to JAX distributed service on mt-13-scan-layers-false-wkj01-slice-job-0-0.mt-13-scan-layers-false-wkj01:8482
I0417 14:09:47.585783 133882125645632 max_utils.py:284] Jax distributed system initialized!
I0417 14:09:52.779340 133882125645632 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 14:09:52.779442 133882125645632 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 14:09:52.779482 133882125645632 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0417 14:09:52.779518 133882125645632 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 14:09:53.456602 133882125645632 maxtext_utils.py:1548] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 14:09:53.456897 133882125645632 checkpointing.py:677] Setting up checkpoint logger...
I0417 14:09:53.456953 133882125645632 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 14:09:53.456998 133882125645632 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 14:09:53.457326 133882125645632 base_pytree_checkpoint_handler.py:441] 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 0x79c325051cd0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 14:09:56.402965 133882125645632 checkpointing.py:265] Enabling policy for fixed interval checkpointing.
I0417 14:09:56.403192 133882125645632 checkpoint_manager.py:708] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79afa46a0b90>}, handler_registry=None
I0417 14:09:56.403428 133882125645632 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79afa46a0b90>` 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`.
I0417 14:09:56.403477 133882125645632 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afa46a0dd0>` 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`.
I0417 14:09:56.403519 133882125645632 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 0x79afa46a0b90>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79afa46a0b90>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afa46a0dd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79afa46a0dd0>}).
I0417 14:09:56.403850 133882125645632 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 14:09:56.403919 133882125645632 async_checkpointer.py:192] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x79afa42d0a40> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 14:09:57.577494 133882125645632 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints
I0417 14:09:58.872760 133882125645632 checkpoint_manager.py:929] [process=6][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_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, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79afa46a2990>
I0417 14:09:58.872969 133882125645632 checkpointing.py:301] Checkpoint manager created!
I0417 14:09:59.848685 133882125645632 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 14:09:59.848791 133882125645632 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 14:10:00.229001 133882125645632 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 14:10:00.229094 133882125645632 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 14:10:00.245480 133882125645632 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 14:10:00.245555 133882125645632 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 14:10:00.275456 133882125645632 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 14:10:00.275537 133882125645632 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 14:10:00.291782 133882125645632 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 14:10:00.291851 133882125645632 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 14:10:00.307933 133882125645632 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 14:10:00.307994 133882125645632 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 14:10:00.337342 133882125645632 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 14:10:00.337408 133882125645632 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 14:10:00.359730 133882125645632 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 14:10:00.359804 133882125645632 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 14:10:03.216854 133882125645632 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 14:10:03.216984 133882125645632 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0417 14:10:10.403972 133882125645632 maxtext_utils.py:1651]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404105 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404158 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404215 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.404256 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404292 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404343 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404392 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.404430 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404466 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_0/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404500 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404534 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404567 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.404597 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404627 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404658 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404692 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.404726 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404757 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_1/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404787 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404837 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.404874 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.404904 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404933 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.404963 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.404993 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.405024 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405053 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_10/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405083 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405112 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405142 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.405170 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405198 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405228 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405258 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.405287 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405317 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_11/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405347 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405377 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405407 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.405447 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405498 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405536 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405569 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.405601 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405641 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_12/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405673 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405704 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.405734 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.405762 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405790 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.405854 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405892 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.405924 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405955 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_13/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.405984 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406013 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406043 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.406070 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406098 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406126 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406157 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.406187 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406216 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_14/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406245 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406274 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406303 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.406331 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406358 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406386 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406416 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.406445 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406475 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_15/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406504 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406534 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406562 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.406589 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406619 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406648 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406679 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.406709 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406738 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_2/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406767 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406796 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.406842 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.406872 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406900 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.406929 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.406959 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.406988 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407017 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_3/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407046 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407074 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407103 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.407130 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407157 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407186 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407216 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.407245 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407274 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_4/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407302 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407330 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407359 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.407386 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407413 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407442 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407470 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.407499 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407528 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_5/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407556 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407584 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407613 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.407640 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407667 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407696 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407725 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.407754 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407782 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_6/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.407815 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407859 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.407889 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.407916 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407943 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.407971 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408000 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.408030 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408058 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_7/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408087 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.408115 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.408144 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.408170 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.408198 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.408226 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408255 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.408283 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408312 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_8/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408340 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.408369 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 14:10:10.408398 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 14:10:10.408424 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.408451 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 14:10:10.408479 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408508 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 14:10:10.408537 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408566 133882125645632 maxtext_utils.py:1651]  params/params/decoder/layers_9/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 14:10:10.408613 133882125645632 maxtext_utils.py:1651]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0417 14:10:10.408657 133882125645632 maxtext_utils.py:1651]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0417 14:10:14.845350 133882125645632 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 14:10:14.845448 133882125645632 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 14:10:14.860895 133882125645632 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 14:10:14.860956 133882125645632 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 14:11:11.177747 133882125645632 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0417 14:11:11.178946 133882125645632 metric_logger.py:301] number parameters: 1.104 billion
I0417 14:12:11.699013 133882125645632 checkpointing.py:772] Waiting for step 0 to finish before checkpoint...
I0417 14:12:11.931576 133882125645632 checkpointing.py:776] Waited 0.2325429916381836 seconds for step 0 to finish before starting checkpointing.
I0417 14:12:11.935030 133882125645632 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 14:12:11.936959 133882125645632 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 0
I0417 14:12:11.938300 133882125645632 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/0.
I0417 14:12:12.724001 133882125645632 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 14:12:12.724977 133882125645632 jax_array_handlers.py:360] Scheduling D2H of 444 prioritized jax.Array.
I0417 14:12:12.725102 133882125645632 replica_slices.py:424] 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
I0417 14:12:13.025957 133882125645632 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.302480s
I0417 14:12:13.026124 133882125645632 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 4.584 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.33652782440185547 s) (per-host)
I0417 14:12:13.026174 133882125645632 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.336588s (batch_requests_ready=0.017071s, total_serialization_initiated=0.319449s, others=0.000068s)
I0417 14:12:13.026426 133882125645632 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.340720s (all_items=0.000020s, per_item={'items': '0.00001955'}, temp_paths=0.340700)
I0417 14:12:13.027253 133882125645632 event_tracking.py:125] [process=6] [async] Finished blocking save in 1.09 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/0.
I0417 14:12:13.027547 133756671768320 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:32:13.027515
I0417 14:12:13.080077 133882125645632 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 14:12:13.080442 133759908034304 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0417 14:12:13.080612 133882125645632 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776435131.9350126, 'wait_for_prev_duration_secs': 6.151199340820312e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776435131.936998, 'checkpointer_blocking_duration_secs': 1.090702772140503, 'get_old_steps_start_time': 1776435133.0277274, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776435131.9331968, 'checkpoint_manager_blocking_duration_secs': 1.1473755836486816}
I0417 14:12:13.080800 133882125645632 checkpointing.py:408] Started an asynchronous checkpoint save for step 0
I0417 14:12:13.080876 133882125645632 max_utils.py:750] 
Memstats: After params initialized:
I0417 14:12:13.080931 133882125645632 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_24(process=6,(0,6,0,0))
I0417 14:12:13.080965 133882125645632 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_25(process=6,(1,6,0,0))
I0417 14:12:13.080993 133882125645632 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_28(process=6,(0,7,0,0))
I0417 14:12:13.081018 133882125645632 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_29(process=6,(1,7,0,0))
I0417 14:12:13.438658 133882125645632 metric_logger.py:196] completed step: 0, seconds: 60.520, TFLOP/s/device: 0.225, Tokens/s/device: 33.840, total_weights: 65536, loss: 10.877, lm_loss: 10.877, perplexity: 52938.617
I0417 14:12:13.574902 133882125645632 metric_logger.py:196] completed step: 1, seconds: 1.730, TFLOP/s/device: 7.852, Tokens/s/device: 1183.548, total_weights: 65536, loss: 10.877, lm_loss: 10.877, perplexity: 52938.617
I0417 14:12:13.990739 133882125645632 metric_logger.py:196] completed step: 2, seconds: 0.015, TFLOP/s/device: 889.094, Tokens/s/device: 134013.873, total_weights: 65536, loss: 10.268, lm_loss: 10.268, perplexity: 28794.377
I0417 14:12:14.126906 133882125645632 metric_logger.py:196] completed step: 3, seconds: 0.410, TFLOP/s/device: 33.108, Tokens/s/device: 4990.375, total_weights: 65536, loss: 9.741, lm_loss: 9.741, perplexity: 16998.465
I0417 14:12:14.403157 133882125645632 metric_logger.py:196] completed step: 4, seconds: 0.143, TFLOP/s/device: 95.285, Tokens/s/device: 14362.355, total_weights: 65536, loss: 9.285, lm_loss: 9.285, perplexity: 10778.518
I0417 14:12:14.414533 133882125645632 metric_logger.py:196] completed step: 5, seconds: 0.136, TFLOP/s/device: 99.922, Tokens/s/device: 15061.260, total_weights: 65536, loss: 8.900, lm_loss: 8.900, perplexity: 7335.202
I0417 14:12:16.414193    2770 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 14:12:18.931079 133756128052992 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/0/items/array_metadatas/process_6
I0417 14:12:39.034192 133882125645632 metric_logger.py:196] completed step: 6, seconds: 0.277, TFLOP/s/device: 49.007, Tokens/s/device: 7386.888, total_weights: 65536, loss: 8.602, lm_loss: 8.602, perplexity: 5440.185
I0417 14:12:39.173880 133882125645632 metric_logger.py:196] completed step: 7, seconds: 24.488, TFLOP/s/device: 0.555, Tokens/s/device: 83.632, total_weights: 65536, loss: 8.393, lm_loss: 8.393, perplexity: 4418.134
I0417 14:12:39.310178 133882125645632 metric_logger.py:196] completed step: 8, seconds: 0.143, TFLOP/s/device: 95.194, Tokens/s/device: 14348.670, total_weights: 65536, loss: 8.264, lm_loss: 8.264, perplexity: 3882.514
I0417 14:12:39.447414 133882125645632 checkpointing.py:772] Waiting for step 9 to finish before checkpoint...
I0417 14:12:39.450716 133882125645632 checkpointing.py:776] Waited 0.003323078155517578 seconds for step 9 to finish before starting checkpointing.
I0417 14:12:39.453689 133882125645632 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 14:12:46.001099 133756671768320 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 47.419 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 33.31145167350769 s) (per-host)
I0417 14:12:46.001224 133756671768320 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 32.973564s.
I0417 14:12:55.426475 133756671768320 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 42.398798s.
I0417 14:12:55.426772 133759908034304 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 14:12:55.426937 133759908034304 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0417 14:12:55.426993 133759908034304 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0417 14:12:55.429321 133759908034304 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0417 14:12:55.429522 133882125645632 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0417 14:12:55.429675 133882125645632 checkpoint_manager.py:1452] Waiting for previous save to complete took 15.975984 seconds. If this number is high, consider checkpointing less frequently.
I0417 14:12:55.432323 133882125645632 checkpoint_manager.py:1512] [process=6] Saving checkpoint at step 9
I0417 14:12:55.434499 133882125645632 event_tracking.py:70] [process=6] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/9.
I0417 14:12:56.168013 133882125645632 jax_array_handlers.py:360] Scheduling D2H of 444 prioritized jax.Array.
I0417 14:12:56.168200 133882125645632 replica_slices.py:424] 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
I0417 14:12:56.295601 133882125645632 base_pytree_checkpoint_handler.py:154] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.128936s
I0417 14:12:56.295768 133882125645632 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/blocking_gbytes_per_sec: 9.551 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.161515474319458 s) (per-host)
I0417 14:12:56.295843 133882125645632 base_pytree_checkpoint_handler.py:768] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.161581s (batch_requests_ready=0.016713s, total_serialization_initiated=0.144795s, others=0.000072s)
I0417 14:12:56.296130 133882125645632 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.166025s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.166008)
I0417 14:12:56.296912 133882125645632 event_tracking.py:125] [process=6] [async] Finished blocking save in 0.86 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/9.
I0417 14:12:56.297284 133759908034304 async_checkpointer.py:76] [process=6][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 14:32:56.297247
I0417 14:12:56.302838 133882125645632 checkpoint_manager.py:1560] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 14:12:56.303134 133756663375616 async_checkpointer.py:280] [process=6][thread=save_finalize] Waiting for background save thread=async_save.
I0417 14:12:56.303296 133882125645632 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776435159.453659, 'wait_for_prev_duration_secs': 15.9759840965271, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776435175.4323633, 'checkpointer_blocking_duration_secs': 0.8650622367858887, 'get_old_steps_start_time': 1776435176.2974505, 'get_old_steps_duration_secs': 3.0517578125e-05, 'checkpoint_manager_blocking_start_time': 1776435159.4516065, 'checkpoint_manager_blocking_duration_secs': 16.851653337478638}
I0417 14:12:56.303508 133882125645632 checkpointing.py:408] Started an asynchronous checkpoint save for step 9
I0417 14:12:56.303556 133882125645632 checkpoint_manager.py:2020] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 14:13:03.961447 133757185009408 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260417_125726/linen_xpk_main_20260417_125726_13_scan_layers_false/checkpoints/9/items/array_metadatas/process_6
I0417 14:13:39.257130 133759908034304 base_pytree_checkpoint_handler.py:130] [process=6] /jax/orbax/write/gbytes_per_sec: 36.630 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.12285113334656 s) (per-host)
I0417 14:13:39.257241 133759908034304 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.959846s.
I0417 14:13:47.301070 133759908034304 async_checkpointer.py:160] [process=6][thread=async_save] Background save thread done. Time taken: 51.003660s.
I0417 14:13:47.301352 133756663375616 async_checkpointer.py:288] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 14:13:47.301474 133756663375616 async_checkpointer.py:298] [process=6][thread=save_finalize] No errors found in background save thread=async_save.
I0417 14:13:47.301525 133756663375616 checkpoint_manager.py:2137] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0417 14:13:47.304095 133756663375616 checkpoint_manager.py:2146] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0417 14:13:47.304272 133882125645632 checkpoint_manager.py:2032] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0417 14:13:47.304417 133882125645632 checkpoint_manager.py:2009] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 14:13:47.305408 133882125645632 metric_logger.py:196] completed step: 9, seconds: 0.139, TFLOP/s/device: 97.790, Tokens/s/device: 14739.963, total_weights: 65536, loss: 8.188, lm_loss: 8.188, perplexity: 3599.028
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Fri Apr 17 14:13:58 UTC 2026
EXIT_CODE=0
XPK Start: Fri Apr 17 03:39:05 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 03:39:29.247146: 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)
I0417 03:39:29.423639 134332767708992 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 03:39:38.463405 134332767708992 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 03:39:38.465681 134332767708992 distributed.py:172] Connecting to JAX distributed service on mt-13-scan-layers-false-m1dnv-slice-job-0-0.mt-13-scan-layers-false-m1dnv:8482
I0417 03:39:39.698132 134332767708992 max_utils.py:284] Jax distributed system initialized!
I0417 03:39:45.853635 134332767708992 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 03:39:45.853737 134332767708992 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 03:39:45.853776 134332767708992 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0417 03:39:45.853809 134332767708992 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 03:39:46.518069 134332767708992 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 03:39:46.518651 134332767708992 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 03:39:46.518827 134332767708992 checkpointing.py:688] Setting up checkpoint logger...
I0417 03:39:46.518892 134332767708992 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 03:39:46.518936 134332767708992 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 03:39:46.519259 134332767708992 base_pytree_checkpoint_handler.py:441] 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 0x7a2c27f141a0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 03:39:49.306088 134332767708992 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0417 03:39:49.306321 134332767708992 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a181418c7d0>}, handler_registry=None
I0417 03:39:49.306563 134332767708992 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a181418c7d0>` 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`.
I0417 03:39:49.306616 134332767708992 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a17e471d580>` 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`.
I0417 03:39:49.306651 134332767708992 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 0x7a181418c7d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7a181418c7d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a17e471d580>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7a17e471d580>}).
I0417 03:39:49.306979 134332767708992 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 03:39:49.307051 134332767708992 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7a17e46a07c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 03:39:50.895222 134332767708992 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints
I0417 03:39:51.747397 134332767708992 checkpoint_manager.py:929] [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_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, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7a17e471fe00>
I0417 03:39:51.747563 134332767708992 checkpointing.py:302] Checkpoint manager created!
I0417 03:39:52.674536 134332767708992 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 03:39:52.674648 134332767708992 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:39:53.055347 134332767708992 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 03:39:53.055435 134332767708992 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:39:53.071639 134332767708992 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 03:39:53.071697 134332767708992 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:39:53.101078 134332767708992 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:39:53.101148 134332767708992 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:39:53.117700 134332767708992 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:39:53.117769 134332767708992 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:39:53.133926 134332767708992 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:39:53.133985 134332767708992 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:39:53.162551 134332767708992 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 03:39:53.162618 134332767708992 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:39:53.185514 134332767708992 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 03:39:53.185578 134332767708992 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 03:39:56.051591 134332767708992 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 03:39:56.051720 134332767708992 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0417 03:40:03.276988 134332767708992 maxtext_utils.py:1796]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277118 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277170 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277227 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.277268 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277305 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277355 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277405 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.277444 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277478 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_0/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277513 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277546 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277578 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.277610 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277641 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277674 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277709 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.277743 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277776 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_1/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.277808 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277854 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.277892 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.277927 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277957 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.277990 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278022 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.278053 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278085 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_10/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278115 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278145 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278175 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.278204 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278232 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278261 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278292 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.278322 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278351 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_11/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278381 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278410 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278439 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.278467 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278495 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278524 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278554 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.278584 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278613 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_12/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278643 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278673 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278702 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.278730 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278757 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.278787 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278816 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.278857 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278889 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_13/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.278923 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278954 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.278983 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.279011 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279039 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279069 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279099 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.279128 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279158 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_14/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279187 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279216 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279245 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.279271 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279299 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279328 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279356 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.279386 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279415 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_15/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279444 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279474 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279503 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.279530 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279558 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279589 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279620 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.279649 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279679 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_2/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279707 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279736 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.279765 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.279792 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279819 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.279856 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279887 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.279920 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279950 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_3/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.279978 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280008 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280036 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.280063 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280090 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280118 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280148 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.280176 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280205 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_4/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280234 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280262 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280291 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.280318 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280345 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280374 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280402 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.280431 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280460 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_5/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280489 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280517 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280546 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.280573 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280600 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280628 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280657 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.280686 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280715 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_6/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280743 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280771 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.280800 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.280827 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280865 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.280895 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280930 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.280961 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.280990 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_7/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281019 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.281048 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.281077 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.281104 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.281131 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.281160 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281189 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.281218 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281247 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_8/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281276 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/mlp/wi_0/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.281304 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/mlp/wi_1/kernel
    Shape:     float32[2048,7168]
    Logical:   P('embed', 'mlp')
    Physical:  ('fsdp', None)
I0417 03:40:03.281333 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/mlp/wo/kernel
    Shape:     float32[7168,2048]
    Logical:   P('mlp', 'embed')
    Physical:  (None, 'fsdp')
I0417 03:40:03.281359 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/post_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.281386 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/pre_self_attention_layer_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:40:03.281414 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/self_attention/key/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281443 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/self_attention/out/kernel
    Shape:     float32[16,128,2048]
    Logical:   P('heads', 'kv', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:40:03.281472 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/self_attention/query/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281500 134332767708992 maxtext_utils.py:1796]  params/params/decoder/layers_9/self_attention/value/kernel
    Shape:     float32[2048,16,128]
    Logical:   P('embed', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None)
I0417 03:40:03.281548 134332767708992 maxtext_utils.py:1796]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)
I0417 03:40:03.281590 134332767708992 maxtext_utils.py:1796]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')

I0417 03:40:07.781395 134332767708992 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 03:40:07.781492 134332767708992 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 03:40:07.797396 134332767708992 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 03:40:07.797461 134332767708992 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 03:41:04.405392 134332767708992 max_utils.py:791] Total memory size: 1.8 GB, Output size: 0.4 GB, Temp size: 1.5 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0417 03:41:04.413709 134332767708992 metric_logger.py:289] number parameters: 1.104 billion
I0417 03:42:06.017291 134332767708992 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0417 03:42:06.264183 134332767708992 checkpointing.py:798] Waited 0.2468705177307129 seconds for step 0 to finish before starting checkpointing.
I0417 03:42:06.267822 134332767708992 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 03:42:06.269573 134332767708992 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0
I0417 03:42:06.270938 134332767708992 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0.
I0417 03:42:06.585407 134332767708992 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 03:42:06.630973 134332767708992 jax_array_handlers.py:360] Scheduling D2H of 444 prioritized jax.Array.
I0417 03:42:06.631114 134332767708992 replica_slices.py:424] 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
I0417 03:42:06.830291 134203895179008 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0
I0417 03:42:06.930945 134332767708992 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.301317s
I0417 03:42:06.931788 134332767708992 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 4.551 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.3389620780944824 s) (per-host)
I0417 03:42:06.931869 134332767708992 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.339061s (batch_requests_ready=0.017997s, total_serialization_initiated=0.320288s, others=0.000776s)
I0417 03:42:06.931987 134332767708992 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.342891s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.342873)
I0417 03:42:06.932691 134332767708992 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.66 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0.
I0417 03:42:06.933054 134204388292352 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:02:06.933016
I0417 03:42:06.971814 134332767708992 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 03:42:06.972132 134203366700800 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 03:42:06.972271 134332767708992 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776397326.267804, 'wait_for_prev_duration_secs': 7.915496826171875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776397326.2696116, 'checkpointer_blocking_duration_secs': 0.6635925769805908, 'get_old_steps_start_time': 1776397326.9332278, 'get_old_steps_duration_secs': 2.8848648071289062e-05, 'checkpoint_manager_blocking_start_time': 1776397326.2657056, 'checkpoint_manager_blocking_duration_secs': 0.7065262794494629}
I0417 03:42:06.972476 134332767708992 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0417 03:42:06.972525 134332767708992 max_utils.py:750] 
Memstats: After params initialized:
I0417 03:42:06.972583 134332767708992 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_0(process=0,(0,0,0,0))
I0417 03:42:06.972615 134332767708992 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_1(process=0,(1,0,0,0))
I0417 03:42:06.972643 134332767708992 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_4(process=0,(0,1,0,0))
I0417 03:42:06.972666 134332767708992 max_utils.py:756] 	Using (GB) 0.82 / 31.25 (2.624000%) on TPU_5(process=0,(1,1,0,0))
I0417 03:42:07.342226 134332767708992 metric_logger.py:185] completed step: 0, seconds: 60.990, TFLOP/s/device: 0.223, Tokens/s/device: 33.579, total_weights: 65536, loss: 10.877
I0417 03:42:07.344944 134332767708992 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/tensorboard/'
I0417 03:42:07.768883 134332767708992 metric_logger.py:185] completed step: 1, seconds: 1.316, TFLOP/s/device: 10.323, Tokens/s/device: 1556.070, total_weights: 65536, loss: 10.877
I0417 03:42:07.900859 134332767708992 metric_logger.py:185] completed step: 2, seconds: 0.433, TFLOP/s/device: 31.392, Tokens/s/device: 4731.726, total_weights: 65536, loss: 10.268
I0417 03:42:08.037302 134332767708992 metric_logger.py:185] completed step: 3, seconds: 0.011, TFLOP/s/device: 1201.763, Tokens/s/device: 181142.756, total_weights: 65536, loss: 9.741
I0417 03:42:09.657574 134203339974400 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776397329255093182, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776397329255093182, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:42:09.934739 134203878393600 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/items
I0417 03:42:11.960443    2893 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 03:42:14.256170 134203861608192 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/items/array_metadatas/process_0
I0417 03:42:32.183052 134332767708992 metric_logger.py:185] completed step: 4, seconds: 0.134, TFLOP/s/device: 101.686, Tokens/s/device: 15327.276, total_weights: 65536, loss: 9.285
I0417 03:42:32.197465 134332767708992 metric_logger.py:185] completed step: 5, seconds: 0.136, TFLOP/s/device: 99.599, Tokens/s/device: 15012.682, total_weights: 65536, loss: 8.900
I0417 03:42:32.327995 134332767708992 metric_logger.py:185] completed step: 6, seconds: 24.146, TFLOP/s/device: 0.563, Tokens/s/device: 84.816, total_weights: 65536, loss: 8.602
I0417 03:42:32.464168 134332767708992 metric_logger.py:185] completed step: 7, seconds: 0.012, TFLOP/s/device: 1150.769, Tokens/s/device: 173456.424, total_weights: 65536, loss: 8.393
I0417 03:42:32.600366 134332767708992 metric_logger.py:185] completed step: 8, seconds: 0.132, TFLOP/s/device: 102.965, Tokens/s/device: 15519.972, total_weights: 65536, loss: 8.264
I0417 03:42:32.736177 134332767708992 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0417 03:42:32.739555 134332767708992 checkpointing.py:798] Waited 0.003393888473510742 seconds for step 9 to finish before starting checkpointing.
I0417 03:42:32.742853 134332767708992 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 03:42:45.402987 134203853215488 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 33.471424s (commit=31.850805s, array_metadata_write=1.620619s)
I0417 03:42:45.404688 134204388292352 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 40.699 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 38.81181502342224 s) (per-host)
I0417 03:42:45.404863 134204388292352 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.471674s.
I0417 03:42:46.086142 134204388292352 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776397329255093182, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:42:47.279818 134203339974400 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776397329255093182, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:42:47.620540 134204388292352 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.002281s.
I0417 03:42:47.622327 134204388292352 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 03:42:48.974731 134204388292352 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.333540s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/items
I0417 03:42:48.976344 134204388292352 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0/items
I0417 03:42:49.513123 134204388292352 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0
I0417 03:42:50.951374 134204388292352 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0`.
I0417 03:42:50.952255 134204388292352 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 44.68 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/0
I0417 03:42:50.953690 134204388292352 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 44.020519s.
I0417 03:42:50.953930 134203366700800 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 03:42:50.954051 134203366700800 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 03:42:50.954119 134203366700800 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0417 03:42:50.956147 134203366700800 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0417 03:42:50.956333 134332767708992 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0.
W0417 03:42:50.956470 134332767708992 checkpoint_manager.py:1452] Waiting for previous save to complete took 18.213630 seconds. If this number is high, consider checkpointing less frequently.
I0417 03:42:50.958950 134332767708992 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9
I0417 03:42:50.960770 134332767708992 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9.
I0417 03:42:51.697121 134332767708992 jax_array_handlers.py:360] Scheduling D2H of 444 prioritized jax.Array.
I0417 03:42:51.697284 134332767708992 replica_slices.py:424] 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
I0417 03:42:51.821686 134332767708992 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.125886s
I0417 03:42:51.822471 134332767708992 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 9.525 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.16194987297058105 s) (per-host)
I0417 03:42:51.822536 134332767708992 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.162031s (batch_requests_ready=0.017427s, total_serialization_initiated=0.143895s, others=0.000709s)
I0417 03:42:51.822653 134332767708992 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.165958s (all_items=0.000016s, per_item={'items': '0.00001645'}, temp_paths=0.165942)
I0417 03:42:51.823397 134332767708992 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.86 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9.
I0417 03:42:51.823749 134203366700800 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 04:02:51.823713
I0417 03:42:51.826945 134332767708992 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 03:42:51.827212 134203853215488 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 03:42:51.827373 134332767708992 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776397352.7428057, 'wait_for_prev_duration_secs': 18.213630437850952, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776397370.9589913, 'checkpointer_blocking_duration_secs': 0.8648688793182373, 'get_old_steps_start_time': 1776397371.8238838, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776397352.7403944, 'checkpoint_manager_blocking_duration_secs': 19.086946725845337}
I0417 03:42:51.827549 134332767708992 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0417 03:42:51.827594 134332767708992 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 03:42:51.910460 134204388292352 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9
I0417 03:42:54.079990 134204921345792 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9/items
I0417 03:42:57.294781 134203895179008 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9/items/array_metadatas/process_0
I0417 03:43:37.103287 134203861608192 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 41.742625s (commit=39.501849s, array_metadata_write=2.240776s)
I0417 03:43:37.105159 134203366700800 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.759 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.44459581375122 s) (per-host)
I0417 03:43:37.105285 134203366700800 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.280759s.
I0417 03:43:38.756941 134203366700800 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.002776s.
I0417 03:43:39.046182 134203366700800 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 03:43:40.281946 134203366700800 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 2.282154s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9/items
I0417 03:43:40.283648 134203366700800 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9/items
I0417 03:43:40.803235 134203366700800 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9
I0417 03:43:42.200947 134203366700800 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9`.
I0417 03:43:42.201785 134203366700800 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 51.24 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_13_scan_layers_false/checkpoints/9
I0417 03:43:42.203402 134203366700800 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.378874s.
I0417 03:43:42.203610 134203853215488 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 03:43:42.203726 134203853215488 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 03:43:42.203790 134203853215488 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0417 03:43:42.205523 134203853215488 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0417 03:43:42.205702 134332767708992 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0417 03:43:42.205863 134332767708992 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 03:43:42.206523 134332767708992 metric_logger.py:185] completed step: 9, seconds: 0.135, TFLOP/s/device: 100.462, Tokens/s/device: 15142.777, total_weights: 65536, loss: 8.188
Per train step:
 Total TFLOPs: 13.59 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Fri Apr 17 03:43:51 UTC 2026
EXIT_CODE=0