main (172d0f10a) vs fix/gate-fp8-nan-sanitization (3d212094d)| Metric | main 172d0f10a | fix/gate-fp8-nan-sanitization 3d212094d | Diff (fix/gate-fp8-nan-sanitization − main) |
|---|---|---|---|
| Parameters | 1.104 billion | 1.104 billion | — |
| Final loss | 5.9300 | 5.9300 | 0 |
| TFLOP/s | 686.080 | 148.137 | -537.943 |
| Tok/s | 103413.5 | 22328.8 | -81084.625 |
| Avg s/step | 3.180 | 2.606 | -0.574 |
| Memory % | 6.85 | 6.85 | 0 |
| JAX | 0.8.1 | 0.8.1 | — |
Diff = branch value − main value. Green = branch improved. Red = branch regressed.
2026-04-23 02:01:26.344730: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776909686.358063 2300539 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776909686.361952 2300539 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776909686.373473 2300539 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909686.373488 2300539 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909686.373490 2300539 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776909686.373491 2300539 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-23 02:01:30.048109: 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) I0423 02:01:30.183031 133867303257920 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. I0423 02:02:01.774880 133867303257920 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 02:02:01.775075 133867303257920 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 02:02:01.775116 133867303257920 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0423 02:02:01.775148 133867303257920 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 02:02:01.775170 133867303257920 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 02:02:01.775280 133867303257920 train.py:703] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 02:02:02.430383 133867303257920 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 02:02:02.430642 133867303257920 checkpointing.py:677] Setting up checkpoint logger... I0423 02:02:02.430686 133867303257920 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 02:02:02.430721 133867303257920 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 02:02:02.431171 133867303257920 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x79c01f7a7350>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 02:02:04.852381 133867303257920 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 02:02:04.852955 133867303257920 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79bec2d63170>}, handler_registry=None I0423 02:02:04.853380 133867303257920 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79bec2d63170>` 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`. I0423 02:02:04.853428 133867303257920 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79bec236c920>` 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`. I0423 02:02:04.853462 133867303257920 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 0x79bec2d63170>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x79bec2d63170>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79bec236c920>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x79bec236c920>}). I0423 02:02:04.854016 133867303257920 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 02:02:04.854119 133867303257920 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x79be547e8860> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 02:02:04.985792 133867303257920 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints I0423 02:02:04.986065 133867303257920 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x79bec236e360> I0423 02:02:04.986356 133867303257920 checkpointing.py:301] Checkpoint manager created! I0423 02:02:05.402097 133867303257920 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 02:02:05.402238 133867303257920 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:02:06.044930 133867303257920 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:02:06.045060 133867303257920 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:02:06.057454 133867303257920 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 02:02:06.057531 133867303257920 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 02:02:06.078786 133867303257920 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:02:06.078871 133867303257920 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:02:06.091385 133867303257920 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:02:06.091466 133867303257920 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:02:06.103853 133867303257920 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 02:02:06.103929 133867303257920 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:02:06.124775 133867303257920 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 02:02:06.124863 133867303257920 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 02:02:06.143637 133867303257920 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 02:02:06.143726 133867303257920 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 02:02:08.568816 133867303257920 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 02:02:08.568936 133867303257920 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0423 02:02:13.897916 133867303257920 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898143 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898188 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898234 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.898264 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898291 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898331 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898370 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.898400 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.898426 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898452 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898478 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898503 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.898527 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898549 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898571 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898595 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.898619 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.898643 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898665 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898687 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898709 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.898729 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898750 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898773 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898794 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.898817 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.898846 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898869 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898890 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.898912 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.898932 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898953 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.898975 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.898997 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.899018 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.899039 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899078 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899099 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899122 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.899145 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899166 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899188 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899210 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.899232 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.899254 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899278 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899299 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899320 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.899341 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899362 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899384 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899406 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.899427 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.899448 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899470 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899491 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899512 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.899532 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899552 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899573 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899594 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.899615 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.899636 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899657 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899680 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899702 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.899723 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899743 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899765 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899786 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.899808 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.899833 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899855 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899878 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.899900 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.899920 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899940 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.899961 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.899984 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900005 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900027 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900059 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900085 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900106 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.900126 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900146 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900169 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900190 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900212 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900233 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900254 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900275 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900296 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.900316 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900336 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900357 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900379 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900400 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900422 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900444 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900465 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900486 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.900505 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900525 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900547 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900567 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900589 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900610 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900630 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900651 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900672 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.900692 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900712 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900734 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900756 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900778 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900799 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900820 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900844 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.900865 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.900885 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900905 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.900928 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.900949 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.900971 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.900992 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.901013 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.901034 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.901071 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.901093 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.901112 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.901133 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.901154 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.901175 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.901197 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.901218 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.901240 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 02:02:13.901261 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 02:02:13.901280 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.901300 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 02:02:13.901321 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.901342 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 02:02:13.901364 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 02:02:13.901386 133867303257920 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 02:02:13.901425 133867303257920 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 02:02:13.901459 133867303257920 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 02:02:16.912881 133867303257920 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:02:16.912974 133867303257920 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:02:16.924562 133867303257920 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 02:02:16.924610 133867303257920 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 02:02:30.031201 133867303257920 max_utils.py:791] Total memory size: 3.3 GB, Output size: 1.5 GB, Temp size: 1.8 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0423 02:02:30.032656 133867303257920 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 02:02:30.033349 133867303257920 metric_logger.py:301] number parameters: 1.104 billion I0423 02:02:59.165156 133867303257920 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 02:02:59.273691 133867303257920 checkpointing.py:776] Waited 0.10860276222229004 seconds for step 0 to finish before starting checkpointing. I0423 02:02:59.276707 133867303257920 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:02:59.277007 133867303257920 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 02:02:59.277896 133867303257920 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0. I0423 02:02:59.375822 133867303257920 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 02:02:59.511327 133836268635712 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0 I0423 02:02:59.512000 133867303257920 jax_array_handlers.py:347] Scheduling D2H of 444 prioritized jax.Array. I0423 02:02:59.512087 133867303257920 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0423 02:03:00.233787 133836258149952 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/items I0423 02:03:00.362905 133867303257920 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.852144s I0423 02:03:00.363398 133867303257920 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.517 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 985 milliseconds) (per-host) I0423 02:03:00.363462 133867303257920 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.986056s (batch_requests_ready=0.105284s, total_serialization_initiated=0.880417s, others=0.000355s) I0423 02:03:00.363604 133867303257920 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.987148s (all_items=0.000047s, per_item={'items': '0.00004745'}, temp_paths=0.987101) I0423 02:03:00.364626 133836033754688 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:03:00.364810 133867303257920 async_checkpointer.py:561] Finished blocking save. Time taken: 1.087748s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0. I0423 02:03:00.365019 133867303257920 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:03:00.365278 133867303257920 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776909779.276684, 'wait_for_prev_duration_secs': 5.936622619628906e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776909779.2770302, 'checkpointer_blocking_duration_secs': 1.087853193283081, 'get_old_steps_start_time': 1776909780.3649116, 'get_old_steps_duration_secs': 5.340576171875e-05, 'checkpoint_manager_blocking_start_time': 1776909779.276432, 'checkpoint_manager_blocking_duration_secs': 1.0888104438781738} I0423 02:03:00.365466 133867303257920 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 02:03:00.365546 133867303257920 max_utils.py:750] Memstats: After params initialized: I0423 02:03:00.365861 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_0(process=0,(0,0,0,0)) I0423 02:03:00.365893 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_1(process=0,(1,0,0,0)) I0423 02:03:00.365913 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_2(process=0,(0,1,0,0)) I0423 02:03:00.365934 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_3(process=0,(1,1,0,0)) I0423 02:03:00.365952 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_4(process=0,(0,2,0,0)) I0423 02:03:00.365971 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_5(process=0,(1,2,0,0)) I0423 02:03:00.365988 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_6(process=0,(0,3,0,0)) I0423 02:03:00.366005 133867303257920 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_7(process=0,(1,3,0,0)) I0423 02:03:00.373661 133836226692672 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909780118505966, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776909780118505966, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 02:03:00.373932 133836279121472 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:03:00.799038 2302660 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 02:03:00.799126 2302660 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 02:03:01.129373 133867303257920 metric_logger.py:196] completed step: 0, seconds: 29.131, TFLOP/s/device: 0.466, Tokens/s/device: 70.304, total_weights: 16384, loss: 10.887, lm_loss: 10.887, perplexity: 53501.250 I0423 02:03:01.132282 133867303257920 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/tensorboard/' I0423 02:03:01.149579 133867303257920 metric_logger.py:196] completed step: 1, seconds: 1.671, TFLOP/s/device: 8.130, Tokens/s/device: 1225.390, total_weights: 16384, loss: 10.887, lm_loss: 10.887, perplexity: 53501.250 I0423 02:03:01.250208 133867303257920 metric_logger.py:196] completed step: 2, seconds: 0.307, TFLOP/s/device: 44.205, Tokens/s/device: 6663.131, total_weights: 16384, loss: 9.787, lm_loss: 9.787, perplexity: 17806.746 I0423 02:03:01.362791 133867303257920 metric_logger.py:196] completed step: 3, seconds: 0.016, TFLOP/s/device: 864.651, Tokens/s/device: 130329.642, total_weights: 16384, loss: 8.853, lm_loss: 8.853, perplexity: 6997.778 I0423 02:03:01.576284 133836044240448 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/items/array_metadatas/process_0 I0423 02:03:23.010005 133867303257920 metric_logger.py:196] completed step: 4, seconds: 0.103, TFLOP/s/device: 131.658, Tokens/s/device: 19844.961, total_weights: 16384, loss: 7.982, lm_loss: 7.982, perplexity: 2927.066 I0423 02:03:27.314468 133867303257920 metric_logger.py:196] completed step: 5, seconds: 0.112, TFLOP/s/device: 121.559, Tokens/s/device: 18322.687, total_weights: 16384, loss: 7.248, lm_loss: 7.248, perplexity: 1405.074 I0423 02:03:27.590990 133867303257920 metric_logger.py:196] completed step: 6, seconds: 24.970, TFLOP/s/device: 0.544, Tokens/s/device: 82.017, total_weights: 16384, loss: 6.690, lm_loss: 6.690, perplexity: 804.397 I0423 02:03:27.776598 133867303257920 metric_logger.py:196] completed step: 7, seconds: 1.123, TFLOP/s/device: 12.098, Tokens/s/device: 1823.532, total_weights: 16384, loss: 6.307, lm_loss: 6.307, perplexity: 548.399 I0423 02:03:27.995761 133867303257920 metric_logger.py:196] completed step: 8, seconds: 0.302, TFLOP/s/device: 45.004, Tokens/s/device: 6783.523, total_weights: 16384, loss: 6.070, lm_loss: 6.070, perplexity: 432.506 I0423 02:03:28.000666 133867303257920 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 02:03:28.014468 133867303257920 checkpointing.py:776] Waited 0.013959169387817383 seconds for step 9 to finish before starting checkpointing. I0423 02:03:28.018759 133867303257920 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:03:54.734252 133836247664192 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 53.975275s (commit=53.474829s, array_metadata_write=0.500446s) I0423 02:03:54.736841 133836033754688 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 228.268 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 55 seconds) (per-host) I0423 02:03:54.736930 133836033754688 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 54.372174s. I0423 02:03:54.993527 133836033754688 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909780118505966, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 02:03:55.200744 133836033754688 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 02:03:55.430428 133836226692672 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776909780118505966, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 02:03:55.591219 133836033754688 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:03:55.592516 133836033754688 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.557426s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/items I0423 02:03:55.593353 133836033754688 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0/items I0423 02:03:55.827105 133836033754688 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0 I0423 02:03:56.543728 133836033754688 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0`. I0423 02:03:56.544523 133836033754688 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 57.267465s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/0 I0423 02:03:56.544629 133836033754688 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 56.179879s. I0423 02:03:56.544861 133836279121472 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:03:56.544928 133836279121472 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:03:56.545008 133836279121472 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:03:56.545077 133836279121472 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 02:03:56.545292 133867303257920 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0423 02:03:56.545475 133867303257920 checkpoint_manager.py:1441] Waiting for previous save to complete took 28.528701 seconds. If this number is high, consider checkpointing less frequently. I0423 02:03:56.549908 133867303257920 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 02:03:56.550410 133867303257920 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9. I0423 02:03:56.726683 133836033754688 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9 I0423 02:03:56.782999 133867303257920 jax_array_handlers.py:347] Scheduling D2H of 444 prioritized jax.Array. I0423 02:03:56.783235 133867303257920 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0423 02:03:57.158039 133867303257920 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.375889s I0423 02:03:57.158720 133867303257920 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 23.819 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 518 milliseconds) (per-host) I0423 02:03:57.158818 133867303257920 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.518245s (batch_requests_ready=0.109390s, total_serialization_initiated=0.408289s, others=0.000566s) I0423 02:03:57.158979 133867303257920 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.519360s (all_items=0.000060s, per_item={'items': '0.00005960'}, temp_paths=0.519300) I0423 02:03:57.159867 133836056823360 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 02:03:57.160004 133867303257920 async_checkpointer.py:561] Finished blocking save. Time taken: 0.610032s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9. I0423 02:03:57.160300 133867303257920 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 02:03:57.160543 133836279121472 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 02:03:57.160751 133867303257920 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776909808.0167234, 'wait_for_prev_duration_secs': 28.528700590133667, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776909836.5499382, 'checkpointer_blocking_duration_secs': 0.6101987361907959, 'get_old_steps_start_time': 1776909837.1601636, 'get_old_steps_duration_secs': 8.7738037109375e-05, 'checkpoint_manager_blocking_start_time': 1776909808.0158195, 'checkpoint_manager_blocking_duration_secs': 29.144875526428223} I0423 02:03:57.160979 133867303257920 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 02:03:57.161018 133867303257920 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 02:03:57.433966 133836044240448 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9/items I0423 02:03:58.730478 133836077794880 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9/items/array_metadatas/process_0 I0423 02:04:41.372424 133836067309120 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 43.449160s (commit=42.962443s, array_metadata_write=0.486717s) I0423 02:04:41.379587 133836056823360 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 282.456 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 44 seconds) (per-host) I0423 02:04:41.379663 133836056823360 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 44.219584s. I0423 02:04:41.828627 133836056823360 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 02:04:42.224465 133836056823360 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 02:04:42.229888 133836056823360 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.570588s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9/items I0423 02:04:42.230834 133836056823360 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9/items I0423 02:04:42.460906 133836056823360 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9 I0423 02:04:43.136366 133836056823360 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9`. I0423 02:04:43.137155 133836056823360 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 46.587189s. directory=gs://wanglance-maxtext/linen_ckpt_main_20260423_012248/linen_main_20260423_012248_13_scan_layers_false/checkpoints/9 I0423 02:04:43.137248 133836056823360 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 45.977176s. I0423 02:04:43.137496 133836279121472 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 02:04:43.137658 133836279121472 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 02:04:43.137757 133836279121472 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 02:04:43.137811 133836279121472 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 02:04:43.138103 133867303257920 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0423 02:04:43.138288 133867303257920 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 02:04:43.139301 133867303257920 metric_logger.py:196] completed step: 9, seconds: 0.020, TFLOP/s/device: 686.080, Tokens/s/device: 103413.452, total_weights: 16384, loss: 5.930, lm_loss: 5.930, perplexity: 376.179 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops
3d212094d
· fix_gate_fp8_nan_sanitization_20260423_024158
· full log
2026-04-23 03:28:59.561202: E external/local_xla/xla/stream_executor/cuda/cuda_fft.cc:467] Unable to register cuFFT factory: Attempting to register factory for plugin cuFFT when one has already been registered WARNING: All log messages before absl::InitializeLog() is called are written to STDERR E0000 00:00:1776914939.575390 2379424 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776914939.579477 2379424 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776914939.592139 2379424 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914939.592154 2379424 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914939.592156 2379424 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776914939.592157 2379424 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} `rope_scaling`'s factor field must be a float >= 1, got 40 `rope_scaling`'s beta_fast field must be a float, got 32 `rope_scaling`'s beta_slow field must be a float, got 1 Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} Unrecognized keys in `rope_scaling` for 'rope_type'='yarn': {'rope_theta'} 2026-04-23 03:29:03.273717: 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) I0423 03:29:03.408924 135240283092800 max_utils.py:238] Skipping jax distributed system due to skip_jax_distributed_system=True flag. ~/maxtext/maxtext_venv/lib/python3.12/site-packages/jax/_src/xla_bridge.py:204: UserWarning: TPU backend initialization is taking more than 60.0 seconds. Did you run your code on all TPU hosts? See https://docs.jax.dev/en/latest/multi_process.html for more information. warnings.warn( I0423 03:30:09.281600 135240283092800 max_utils.py:800] System Information: Jax Version: 0.8.1 I0423 03:30:09.281854 135240283092800 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0423 03:30:09.281906 135240283092800 max_utils.py:802] System Information: Jax Backend: PJRT C API TFRT TPU v6 lite Built on Nov 12 2025 14:16:36 (1762985796) cl/831091709 I0423 03:30:09.281940 135240283092800 train_utils.py:348] WARNING: 'dataset_path' might be pointing your local file system I0423 03:30:09.281963 135240283092800 train_utils.py:361] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0423 03:30:09.282137 135240283092800 train.py:672] [DECOUPLED NO-OP] skipping cloud diagnostics wrapper. I0423 03:30:09.947524 135240283092800 maxtext_utils.py:1604] Num_devices: 8, shape (1, 1, 1, 8, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0423 03:30:09.947793 135240283092800 checkpointing.py:677] Setting up checkpoint logger... I0423 03:30:09.947838 135240283092800 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0423 03:30:09.947874 135240283092800 pytree_checkpoint_handler.py:577] save_device_host_concurrent_bytes=None I0423 03:30:09.948318 135240283092800 base_pytree_checkpoint_handler.py:411] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7affcb7bdac0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0423 03:30:12.365921 135240283092800 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0423 03:30:12.366284 135240283092800 checkpoint_manager.py:702] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afdf2b2a330>}, handler_registry=None I0423 03:30:12.366508 135240283092800 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afdf2b2a330>` 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`. I0423 03:30:12.366549 135240283092800 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afdf2b2acc0>` 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`. I0423 03:30:12.366579 135240283092800 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 0x7afdf2b2a330>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7afdf2b2a330>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afdf2b2acc0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7afdf2b2acc0>}). I0423 03:30:12.366862 135240283092800 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.28 I0423 03:30:12.366911 135240283092800 async_checkpointer.py:177] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>.<lambda> at 0x7afdf2b5c540> timeout: 600 secs and primary_host=0 for async checkpoint writes I0423 03:30:12.510362 135240283092800 checkpoint_manager.py:1788] Found 0 checkpoint steps in gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints I0423 03:30:12.510572 135240283092800 checkpoint_manager.py:921] [process=0][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_hns=False, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False), root_directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7afdf2b2ab40> I0423 03:30:12.510848 135240283092800 checkpointing.py:301] Checkpoint manager created! I0423 03:30:12.920967 135240283092800 nnx_wrappers.py:437] Unknown Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0423 03:30:12.921104 135240283092800 nnx_wrappers.py:437] Unknown Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:30:13.556857 135240283092800 attentions.py:1088] attentions/inputs_q Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:30:13.556966 135240283092800 attentions.py:1088] attentions/inputs_q Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:30:13.569605 135240283092800 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[8,2048,2048]....................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0423 03:30:13.569681 135240283092800 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[8,2048,2048]....................................... ('fsdp', None, None). I0423 03:30:13.593469 135240283092800 attentions.py:1154] attentions/query Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 03:30:13.593551 135240283092800 attentions.py:1154] attentions/query Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:30:13.605923 135240283092800 attentions.py:1155] attentions/key Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 03:30:13.605999 135240283092800 attentions.py:1155] attentions/key Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:30:13.618355 135240283092800 attentions.py:1156] attentions/value Logical: bfloat16[8,2048,16,128]..................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0423 03:30:13.618431 135240283092800 attentions.py:1156] attentions/value Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:30:13.638360 135240283092800 attentions.py:1198] attentions/out Logical: bfloat16[8,2048,16,128]..................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0423 03:30:13.638440 135240283092800 attentions.py:1198] attentions/out Physical: bfloat16[8,2048,16,128]..................................... ('fsdp', None, None, None). I0423 03:30:13.656006 135240283092800 linears.py:525] linears/x Logical: bfloat16[8,2048,7168]....................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0423 03:30:13.656100 135240283092800 linears.py:525] linears/x Physical: bfloat16[8,2048,7168]....................................... ('fsdp', None, None). I0423 03:30:16.077232 135240283092800 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0423 03:30:16.077336 135240283092800 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. [DECOUPLED NO-OP] gcs_storage: using stubs. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] mldiagnostics: using stub. [DECOUPLED NO-OP] workload_monitor: using stub. [DECOUPLED NO-OP] vertex_tensorboard: using stub. fsdp: 8 I0423 03:30:21.681270 135240283092800 maxtext_utils.py:1707] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681393 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681436 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681480 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.681509 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681535 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681574 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.681620 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.681649 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.681674 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_0/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.681698 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681723 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681746 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.681770 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681792 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681816 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.681839 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.681862 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.681885 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_1/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.681908 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681930 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.681952 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.681973 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.681994 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682017 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682039 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.682075 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.682098 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_10/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682121 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682143 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682166 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.682186 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682206 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682228 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682251 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.682275 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.682298 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_11/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682320 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682342 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682364 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.682386 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682406 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682429 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682451 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.682474 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.682496 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_12/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682518 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682539 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682560 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.682580 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682606 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682629 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682651 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.682673 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.682695 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_13/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682716 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682738 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682759 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.682779 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682799 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682821 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682842 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.682864 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.682885 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_14/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.682907 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682928 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.682949 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.682969 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.682989 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683011 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683033 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.683068 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.683091 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_15/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683112 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683135 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683157 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.683178 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683197 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683219 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683241 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.683263 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.683284 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_2/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683306 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683327 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683348 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.683368 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683389 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683410 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683432 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.683453 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.683474 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_3/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683495 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683516 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683537 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.683557 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683576 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683601 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683624 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.683647 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.683669 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_4/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683691 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683712 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683733 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.683753 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683773 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683794 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683816 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.683837 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.683858 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_5/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.683879 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683900 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.683921 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.683941 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683960 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.683982 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684003 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.684024 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.684054 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_6/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684077 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684100 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684121 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.684141 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684160 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684182 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684203 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.684225 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.684247 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_7/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684267 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684288 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684309 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.684330 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684349 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684371 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684392 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.684414 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.684435 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_8/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684456 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wi_0/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684476 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wi_1/kernel Shape: float32[2048,7168] Logical: PartitionSpec('embed', 'mlp') Physical: ('fsdp', None) I0423 03:30:21.684497 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/mlp/wo/kernel Shape: float32[7168,2048] Logical: PartitionSpec('mlp', 'embed') Physical: (None, 'fsdp') I0423 03:30:21.684517 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/post_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684537 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/pre_self_attention_layer_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0423 03:30:21.684559 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/key/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684581 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/out/kernel Shape: float32[16,128,2048] Logical: PartitionSpec('heads', 'kv', 'embed') Physical: (None, None, 'fsdp') I0423 03:30:21.684605 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/query/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'q_heads', 'kv') Physical: ('fsdp', None, None) I0423 03:30:21.684627 135240283092800 maxtext_utils.py:1707] params/params/decoder/layers_9/self_attention/value/kernel Shape: float32[2048,16,128] Logical: PartitionSpec('embed', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None) I0423 03:30:21.684663 135240283092800 maxtext_utils.py:1707] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed_vocab', 'vocab') Physical: ('fsdp', None) I0423 03:30:21.684694 135240283092800 maxtext_utils.py:1707] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0423 03:30:24.612946 135240283092800 train.py:155] train/xent Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:30:24.613026 135240283092800 train.py:155] train/xent Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:30:24.624706 135240283092800 train.py:162] train/z_loss Logical: float32[8,2048]............................................. ('activation_embed_and_logits_batch', 'activation_length'). I0423 03:30:24.624752 135240283092800 train.py:162] train/z_loss Physical: float32[8,2048]............................................. ('fsdp', None). I0423 03:30:37.860936 135240283092800 max_utils.py:791] Total memory size: 3.3 GB, Output size: 1.5 GB, Temp size: 1.8 GB, Argument size: 1.5 GB, Host temp size: 0.0 GB. I0423 03:30:37.862465 135240283092800 max_utils.py:194] tensorboardX not available; using no-op SummaryWriter. I0423 03:30:37.863367 135240283092800 metric_logger.py:301] number parameters: 1.104 billion I0423 03:30:57.617115 135240283092800 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0423 03:30:57.730007 135240283092800 checkpointing.py:776] Waited 0.11294722557067871 seconds for step 0 to finish before starting checkpointing. I0423 03:30:57.733262 135240283092800 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:30:57.733566 135240283092800 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 0 I0423 03:30:57.734400 135240283092800 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0. I0423 03:30:57.831953 135240283092800 signaling_client.py:373] Using ThreadSafeKeyValueSignalingClient I0423 03:30:57.917513 135209351644736 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0 I0423 03:30:57.954827 135240283092800 jax_array_handlers.py:347] Scheduling D2H of 444 prioritized jax.Array. I0423 03:30:57.954987 135240283092800 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0423 03:30:58.642120 135209341158976 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/items I0423 03:30:58.805922 135208957380160 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915058542157285, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776915058542157285, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:30:58.806411 135240283092800 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.852290s I0423 03:30:58.807032 135240283092800 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 12.678 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 973 milliseconds) (per-host) I0423 03:30:58.807146 135240283092800 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.973715s (batch_requests_ready=0.092738s, total_serialization_initiated=0.880482s, others=0.000494s) I0423 03:30:58.807304 135240283092800 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.974831s (all_items=0.000046s, per_item={'items': '0.00004625'}, temp_paths=0.974785) I0423 03:30:58.808706 135209362130496 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:30:58.808855 135240283092800 async_checkpointer.py:561] Finished blocking save. Time taken: 1.075233s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0. I0423 03:30:58.809242 135240283092800 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:30:58.809713 135208936408640 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:30:58.809898 135240283092800 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776915057.7332382, 'wait_for_prev_duration_secs': 6.151199340820312e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776915057.733589, 'checkpointer_blocking_duration_secs': 1.0754361152648926, 'get_old_steps_start_time': 1776915058.8090706, 'get_old_steps_duration_secs': 0.0001049041748046875, 'checkpoint_manager_blocking_start_time': 1776915057.732988, 'checkpoint_manager_blocking_duration_secs': 1.0768463611602783} I0423 03:30:58.810226 135240283092800 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0423 03:30:58.810364 135240283092800 max_utils.py:750] Memstats: After params initialized: I0423 03:30:58.810732 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_0(process=0,(0,0,0,0)) I0423 03:30:58.810770 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_1(process=0,(1,0,0,0)) I0423 03:30:58.810792 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_2(process=0,(0,1,0,0)) I0423 03:30:58.810813 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_3(process=0,(1,1,0,0)) I0423 03:30:58.810832 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_4(process=0,(0,2,0,0)) I0423 03:30:58.810851 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_5(process=0,(1,2,0,0)) I0423 03:30:58.810869 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_6(process=0,(0,3,0,0)) I0423 03:30:58.810886 135240283092800 max_utils.py:756] Using (GB) 2.14 / 31.25 (6.848000%) on TPU_7(process=0,(1,3,0,0)) I0423 03:30:59.244997 2381676 google_auth_provider.cc:149] Using credentials at ~/.config/gcloud/application_default_credentials.json I0423 03:30:59.245069 2381676 google_auth_provider.cc:156] Using OAuth2 AuthProvider I0423 03:30:59.584777 135240283092800 metric_logger.py:196] completed step: 0, seconds: 19.753, TFLOP/s/device: 0.688, Tokens/s/device: 103.682, total_weights: 16384, loss: 10.887, lm_loss: 10.887, perplexity: 53501.250 I0423 03:30:59.585964 135240283092800 metric_logger.py:281] To see full metrics 'tensorboard --logdir=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/tensorboard/' I0423 03:30:59.603871 135240283092800 metric_logger.py:196] completed step: 1, seconds: 1.665, TFLOP/s/device: 8.160, Tokens/s/device: 1230.025, total_weights: 16384, loss: 10.887, lm_loss: 10.887, perplexity: 53501.250 I0423 03:30:59.704168 135240283092800 metric_logger.py:196] completed step: 2, seconds: 0.315, TFLOP/s/device: 43.098, Tokens/s/device: 6496.143, total_weights: 16384, loss: 9.787, lm_loss: 9.787, perplexity: 17806.746 I0423 03:30:59.814101 135240283092800 metric_logger.py:196] completed step: 3, seconds: 0.016, TFLOP/s/device: 830.408, Tokens/s/device: 125168.072, total_weights: 16384, loss: 8.853, lm_loss: 8.853, perplexity: 6997.778 I0423 03:31:20.130712 135208946894400 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/items/array_metadatas/process_0 I0423 03:31:20.136165 135240283092800 metric_logger.py:196] completed step: 4, seconds: 0.102, TFLOP/s/device: 132.610, Tokens/s/device: 19988.483, total_weights: 16384, loss: 7.982, lm_loss: 7.982, perplexity: 2927.066 I0423 03:31:20.612275 135240283092800 metric_logger.py:196] completed step: 5, seconds: 0.112, TFLOP/s/device: 121.158, Tokens/s/device: 18262.234, total_weights: 16384, loss: 7.248, lm_loss: 7.248, perplexity: 1405.074 I0423 03:31:20.860131 135240283092800 metric_logger.py:196] completed step: 6, seconds: 20.562, TFLOP/s/device: 0.661, Tokens/s/device: 99.600, total_weights: 16384, loss: 6.690, lm_loss: 6.690, perplexity: 804.397 I0423 03:31:21.055894 135240283092800 metric_logger.py:196] completed step: 7, seconds: 0.340, TFLOP/s/device: 39.995, Tokens/s/device: 6028.530, total_weights: 16384, loss: 6.307, lm_loss: 6.307, perplexity: 548.399 I0423 03:31:21.079655 135240283092800 metric_logger.py:196] completed step: 8, seconds: 0.250, TFLOP/s/device: 54.347, Tokens/s/device: 8191.771, total_weights: 16384, loss: 6.070, lm_loss: 6.070, perplexity: 432.506 I0423 03:31:21.188526 135240283092800 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0423 03:31:21.196699 135240283092800 checkpointing.py:776] Waited 0.008208036422729492 seconds for step 9 to finish before starting checkpointing. I0423 03:31:21.197839 135240283092800 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:31:55.281675 135209192261184 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 56.077443s (commit=55.559247s, array_metadata_write=0.518195s) I0423 03:31:55.293277 135209362130496 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 219.925 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 57 seconds) (per-host) I0423 03:31:55.293426 135209362130496 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 56.484509s. I0423 03:31:55.655887 135209362130496 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915058542157285, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:31:55.864270 135209362130496 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 03:31:56.118229 135208957380160 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776915058542157285, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/_CHECKPOINT_METADATA I0423 03:31:56.301249 135209362130496 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:31:56.302579 135209362130496 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.603439s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/items I0423 03:31:56.303484 135209362130496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0/items I0423 03:31:56.560143 135209362130496 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0 I0423 03:31:57.289621 135209362130496 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0`. I0423 03:31:57.290461 135209362130496 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 59.556851s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/0 I0423 03:31:57.290554 135209362130496 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 58.481636s. I0423 03:31:57.290838 135208936408640 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:31:57.290987 135208936408640 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:31:57.291100 135208936408640 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:31:57.291173 135208936408640 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0423 03:31:57.291429 135240283092800 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0423 03:31:57.291575 135240283092800 checkpoint_manager.py:1441] Waiting for previous save to complete took 36.093774 seconds. If this number is high, consider checkpointing less frequently. I0423 03:31:57.293308 135240283092800 checkpoint_manager.py:1501] [process=0] Saving checkpoint at step 9 I0423 03:31:57.293655 135240283092800 async_checkpointer.py:452] [process=0] Started async saving checkpoint to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9. I0423 03:31:57.469730 135209362130496 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9 I0423 03:31:57.514803 135240283092800 jax_array_handlers.py:347] Scheduling D2H of 444 prioritized jax.Array. I0423 03:31:57.514970 135240283092800 replica_slices.py:410] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False I0423 03:31:57.991465 135240283092800 base_pytree_checkpoint_handler.py:153] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.477370s I0423 03:31:57.992091 135240283092800 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/blocking_gbytes_per_sec: 20.189 GiB/s (total gbytes: 12.3 GiB) (time elapsed: 611 milliseconds) (per-host) I0423 03:31:57.992158 135240283092800 base_pytree_checkpoint_handler.py:732] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.611373s (batch_requests_ready=0.105194s, total_serialization_initiated=0.505689s, others=0.000489s) I0423 03:31:57.992290 135240283092800 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.611994s (all_items=0.000015s, per_item={'items': '0.00001478'}, temp_paths=0.611979) I0423 03:31:57.993649 135209156609600 async_checkpointer.py:79] [process=0][thread=async_save] Background save thread started. I0423 03:31:57.993777 135240283092800 async_checkpointer.py:561] Finished blocking save. Time taken: 0.700419s. Continuing background save to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9. I0423 03:31:57.993997 135240283092800 checkpoint_manager.py:1549] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0423 03:31:57.994354 135208936408640 async_checkpointer.py:265] [process=0][thread=save_finalize] Waiting for background save thread=async_save. I0423 03:31:57.994525 135240283092800 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776915081.1977699, 'wait_for_prev_duration_secs': 36.09377384185791, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776915117.2933323, 'checkpointer_blocking_duration_secs': 0.7005589008331299, 'get_old_steps_start_time': 1776915117.9939215, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776915081.1977048, 'checkpoint_manager_blocking_duration_secs': 36.79677748680115} I0423 03:31:57.994729 135240283092800 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0423 03:31:57.994768 135240283092800 checkpoint_manager.py:1994] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0423 03:31:58.161038 135208946894400 atomicity.py:137] Creating tmp directory gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9/items I0423 03:32:00.265735 135209181775424 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 444 array_metadata.ArrayMetadata to gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9/items/array_metadatas/process_0 I0423 03:32:35.333560 135209167095360 base_pytree_checkpoint_handler.py:1217] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 36.688114s (commit=36.162392s, array_metadata_write=0.525722s) I0423 03:32:35.339869 135209156609600 base_pytree_checkpoint_handler.py:128] [process=0] /jax/checkpoint/write/gbytes_per_sec: 332.906 MiB/s (total gbytes: 12.3 GiB) (time elapsed: 37 seconds) (per-host) I0423 03:32:35.340016 135209156609600 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 37.346182s. I0423 03:32:35.825901 135209156609600 array_metadata_store.py:367] [process=0][thread=async_save] Skipped cross-host ArrayMetadata validation because only one process is found: process_index=0. I0423 03:32:36.253894 135209156609600 ocdbt_utils.py:56] Param validation support for Zarr3 will be added later (b/362328389). I0423 03:32:36.255198 135209156609600 base_pytree_checkpoint_handler.py:1342] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 0.600448s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9/items I0423 03:32:36.256344 135209156609600 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9/items I0423 03:32:36.509428 135209156609600 atomicity.py:608] Finalizing gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9 I0423 03:32:37.245703 135209156609600 atomicity.py:794] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9`. I0423 03:32:37.246547 135209156609600 async_checkpointer.py:420] Finished async_save (blocking + background). Time taken: 39.953199s. directory=gs://wanglance-maxtext/linen_ckpt_fix_gate_fp8_nan_sanitization_20260423_024158/linen_fix_gate_fp8_nan_sanitization_20260423_024158_13_scan_layers_false/checkpoints/9 I0423 03:32:37.246641 135209156609600 async_checkpointer.py:144] [process=0][thread=async_save] Background save thread done. Time taken: 39.252813s. I0423 03:32:37.246963 135208936408640 async_checkpointer.py:273] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save. I0423 03:32:37.247129 135208936408640 async_checkpointer.py:283] [process=0][thread=save_finalize] No errors found in background save thread=async_save. I0423 03:32:37.247240 135208936408640 checkpoint_manager.py:2103] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0423 03:32:37.247308 135208936408640 checkpoint_manager.py:2112] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0423 03:32:37.247535 135240283092800 checkpoint_manager.py:2006] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0423 03:32:37.247698 135240283092800 checkpoint_manager.py:1983] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0423 03:32:37.248716 135240283092800 metric_logger.py:196] completed step: 9, seconds: 0.092, TFLOP/s/device: 148.137, Tokens/s/device: 22328.827, total_weights: 16384, loss: 5.930, lm_loss: 5.930, perplexity: 376.179 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops