MaxView

XPK Start: Fri Apr 17 02:59:17 UTC 2026
PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used.
2026-04-17 02:59:41.795973: E external/local_xla/xla/stream_executor/cuda/cuda_platform.cc:51] failed call to cuInit: INTERNAL: CUDA error: Failed call to cuInit: UNKNOWN ERROR (303)
I0417 02:59:41.972342 136856905627456 max_utils.py:273] Attempting to initialize the jax distributed system...
I0417 02:59:51.011619 136856905627456 distributed.py:149] Starting JAX distributed service on [::]:8482
I0417 02:59:51.013863 136856905627456 distributed.py:172] Connecting to JAX distributed service on mt-06-grad-accum-0m8eo-slice-job-0-0.mt-06-grad-accum-0m8eo:8482
I0417 02:59:52.492638 136856905627456 max_utils.py:284] Jax distributed system initialized!
I0417 02:59:58.945405 136856905627456 max_utils.py:800] System Information: Jax Version: 0.9.2
I0417 02:59:58.945506 136856905627456 max_utils.py:801] System Information: Jaxlib Version: 0.9.2
I0417 02:59:58.945544 136856905627456 max_utils.py:802] System Information: Jax Backend: PJRT C API
TFRT TPU v6 lite
Built on Mar 4 2026 11:32:08 (1772652728) cl/878335365
I0417 02:59:58.945575 136856905627456 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing.
I0417 02:59:59.619465 136856905627456 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 02:59:59.620054 136856905627456 maxtext_utils.py:1687] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1)
I0417 02:59:59.620230 136856905627456 checkpointing.py:688] Setting up checkpoint logger...
I0417 02:59:59.620278 136856905627456 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True
I0417 02:59:59.620318 136856905627456 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None
I0417 02:59:59.620639 136856905627456 base_pytree_checkpoint_handler.py:441] Created BasePyTreeCheckpointHandler: use_ocdbt=True, use_zarr3=True, pytree_metadata_options=PyTreeMetadataOptions(support_rich_types=False), array_metadata_store=<orbax.checkpoint._src.metadata.array_metadata_store.Store object at 0x7c77dfb18650>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB)
I0417 03:00:02.973700 136856905627456 checkpointing.py:266] Enabling policy for fixed interval checkpointing.
I0417 03:00:02.974063 136856905627456 checkpoint_manager.py:708] [process=0][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c63785bbc80>}, handler_registry=None
I0417 03:00:02.974293 136856905627456 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c63785bbc80>` for item "items" and save args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>` to `_handler_registry`.
I0417 03:00:02.974340 136856905627456 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c63781b2bd0>` for item "metrics" and save args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>` and restore args `<class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>` to `_handler_registry`.
I0417 03:00:02.974374 136856905627456 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 0x7c63785bbc80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7c63785bbc80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c63781b2bd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7c63781b2bd0>}).
I0417 03:00:02.974686 136856905627456 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34
I0417 03:00:02.974755 136856905627456 async_checkpointer.py:192] [process=0][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7c629059c7c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes
I0417 03:00:04.518270 136856905627456 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints
I0417 03:00:05.366965 136856905627456 checkpoint_manager.py:929] [process=0][thread=MainThread] CheckpointManager created,  primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7c63781b1730>
I0417 03:00:05.367136 136856905627456 checkpointing.py:302] Checkpoint manager created!
I0417 03:00:07.061755 136856905627456 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed').
I0417 03:00:07.061878 136856905627456 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:00:07.439625 136856905627456 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 03:00:07.439714 136856905627456 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:00:07.455968 136856905627456 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed').
I0417 03:00:07.456025 136856905627456 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None).
I0417 03:00:07.479464 136856905627456 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:00:07.479535 136856905627456 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:00:07.495758 136856905627456 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:00:07.495816 136856905627456 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:00:07.512034 136856905627456 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim').
I0417 03:00:07.512093 136856905627456 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:00:07.536336 136856905627456 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv').
I0417 03:00:07.536404 136856905627456 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None).
I0417 03:00:07.557691 136856905627456 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp').
I0417 03:00:07.557760 136856905627456 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None).
I0417 03:00:07.767704 136856905627456 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint
I0417 03:00:07.767807 136856905627456 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint.
fsdp: 32
I0417 03:00:09.169274 136856905627456 maxtext_utils.py:1796]  params/params/decoder/decoder_norm/scale
    Shape:     float32[2048]
    Logical:   P('norm',)
    Physical:  (None,)
I0417 03:00:09.169397 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/mlp/wi_0/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 03:00:09.169449 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/mlp/wi_1/kernel
    Shape:     float32[2048,16,7168]
    Logical:   P('embed', 'layers', 'mlp')
    Physical:  ('fsdp', None, None)
I0417 03:00:09.169506 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/mlp/wo/kernel
    Shape:     float32[7168,16,2048]
    Logical:   P('mlp', 'layers', 'embed')
    Physical:  (None, None, 'fsdp')
I0417 03:00:09.169565 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/post_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 03:00:09.169604 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/pre_self_attention_layer_norm/scale
    Shape:     float32[2048,16]
    Logical:   P('norm', 'layers')
    Physical:  (None, None)
I0417 03:00:09.169655 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/self_attention/key/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0417 03:00:09.169705 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/self_attention/out/kernel
    Shape:     float32[16,16,128,2048]
    Logical:   P('heads', 'layers', 'kv', 'embed')
    Physical:  (None, None, None, 'fsdp')
I0417 03:00:09.169745 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/self_attention/query/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'q_heads', 'kv')
    Physical:  ('fsdp', None, None, None)
I0417 03:00:09.169779 136856905627456 maxtext_utils.py:1796]  params/params/decoder/layers/self_attention/value/kernel
    Shape:     float32[2048,16,16,128]
    Logical:   P('embed', 'layers', 'kv_heads', 'kv_head_dim')
    Physical:  ('fsdp', None, None, None)
I0417 03:00:09.169825 136856905627456 maxtext_utils.py:1796]  params/params/decoder/logits_dense/kernel
    Shape:     float32[2048,32000]
    Logical:   P('embed_vocab', 'vocab')
    Physical:  ('fsdp', None)

I0417 03:00:09.169910 136856905627456 maxtext_utils.py:1796]  params/params/token_embedder/embedding
    Shape:     float32[32000,2048]
    Logical:   P('vocab', 'embed_vocab')
    Physical:  (None, 'fsdp')
I0417 03:00:09.192657 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048]............................................... Unknown.
I0417 03:00:09.192720 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048]............................................... (None,).
I0417 03:00:09.207358 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,7168]....................................... Unknown.
I0417 03:00:09.207416 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,7168]....................................... ('fsdp', None, None).
I0417 03:00:09.236652 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[7168,16,2048]....................................... Unknown.
I0417 03:00:09.236722 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[7168,16,2048]....................................... (None, None, 'fsdp').
I0417 03:00:09.251450 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16]............................................ Unknown.
I0417 03:00:09.251506 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16]............................................ (None, None).
I0417 03:00:09.280729 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,16,16,128]..................................... Unknown.
I0417 03:00:09.280788 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,16,16,128]..................................... ('fsdp', None, None, None).
I0417 03:00:09.295429 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[16,16,128,2048]..................................... Unknown.
I0417 03:00:09.295486 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[16,16,128,2048]..................................... (None, None, None, 'fsdp').
I0417 03:00:09.338935 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[2048,32000]......................................... Unknown.
I0417 03:00:09.339007 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[2048,32000]......................................... ('fsdp', None).
I0417 03:00:09.353595 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Logical: float32[32000,2048]......................................... Unknown.
I0417 03:00:09.353650 136856905627456 gradient_accumulation.py:70] gradient_accumulation/inputs Physical: float32[32000,2048]......................................... (None, 'fsdp').
I0417 03:00:10.028962 136856905627456 train.py:157] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 03:00:10.029054 136856905627456 train.py:157] train/xent Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 03:00:10.044285 136856905627456 train.py:164] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length').
I0417 03:00:10.044342 136856905627456 train.py:164] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None).
I0417 03:00:21.538857 136856905627456 max_utils.py:791] Total memory size: 1.7 GB, Output size: 0.4 GB, Temp size: 1.3 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB.
I0417 03:00:21.546253 136856905627456 metric_logger.py:289] number parameters: 1.104 billion
I0417 03:00:23.684207 136856905627456 checkpointing.py:794] Waiting for step 0 to finish before checkpoint...
I0417 03:00:34.071921 136856905627456 checkpointing.py:798] Waited 10.3876953125 seconds for step 0 to finish before starting checkpointing.
I0417 03:00:34.074326 136856905627456 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 03:00:34.076324 136856905627456 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 0
I0417 03:00:34.077716 136856905627456 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0.
I0417 03:00:34.799746 136856905627456 signaling_client.py:364] Using JaxDistributedSignalingClient
I0417 03:00:34.814117 136856905627456 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 03:00:34.814185 136856905627456 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0417 03:00:35.081015 136856905627456 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.267774s
I0417 03:00:35.081804 136856905627456 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 5.632 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2738926410675049 s) (per-host)
I0417 03:00:35.081890 136856905627456 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.273997s (batch_requests_ready=0.002076s, total_serialization_initiated=0.271212s, others=0.000709s)
I0417 03:00:35.081970 136856905627456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.278062s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.278046)
I0417 03:00:35.082763 136856905627456 event_tracking.py:125] [process=0] [async] Finished blocking save in 1.01 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0.
I0417 03:00:35.083113 136727733729024 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 03:20:35.083076
I0417 03:00:35.092174 136727725336320 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0
I0417 03:00:35.101005 136856905627456 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 03:00:35.101279 136725040981760 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 03:00:35.101416 136856905627456 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776394834.0743089, 'wait_for_prev_duration_secs': 6.008148193359375e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776394834.0763605, 'checkpointer_blocking_duration_secs': 1.00689697265625, 'get_old_steps_start_time': 1776394835.0832832, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776394834.0724502, 'checkpoint_manager_blocking_duration_secs': 1.0289256572723389}
I0417 03:00:35.101522 136856905627456 checkpointing.py:409] Started an asynchronous checkpoint save for step 0
I0417 03:00:35.101571 136856905627456 max_utils.py:750] 
Memstats: After params initialized:
I0417 03:00:35.101623 136856905627456 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_0(process=0,(0,0,0,0))
I0417 03:00:35.101654 136856905627456 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_1(process=0,(1,0,0,0))
I0417 03:00:35.101681 136856905627456 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_4(process=0,(0,1,0,0))
I0417 03:00:35.101704 136856905627456 max_utils.py:756] 	Using (GB) 0.43 / 31.25 (1.376000%) on TPU_5(process=0,(1,1,0,0))
I0417 03:00:35.443280 136856905627456 metric_logger.py:185] completed step: 0, seconds: 1.806, TFLOP/s/device: 30.088, Tokens/s/device: 4535.165, total_weights: 262144, loss: 10.877
I0417 03:00:35.446230 136856905627456 metric_logger.py:269] To see full metrics 'tensorboard --logdir=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/tensorboard/'
I0417 03:00:36.019278 136856905627456 metric_logger.py:185] completed step: 1, seconds: 11.757, TFLOP/s/device: 4.622, Tokens/s/device: 696.748, total_weights: 262144, loss: 10.877
I0417 03:00:36.597661 136856905627456 metric_logger.py:185] completed step: 2, seconds: 0.396, TFLOP/s/device: 137.124, Tokens/s/device: 20668.757, total_weights: 262144, loss: 10.563
I0417 03:00:36.928308 136727700158208 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/items
I0417 03:00:37.092022 136722335655680 checkpoint.py:188] Wrote Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776394836675496890, 'commit_timestamp_nsecs': None, 'custom_metadata': {}}, json={"item_handlers": null, "metrics": {}, "performance_metrics": {}, "init_timestamp_nsecs": 1776394836675496890, "commit_timestamp_nsecs": null, "custom_metadata": {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:00:37.175873 136856905627456 metric_logger.py:185] completed step: 3, seconds: 0.190, TFLOP/s/device: 285.544, Tokens/s/device: 43040.356, total_weights: 262144, loss: 10.272
I0417 03:00:38.457180    2676 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com
I0417 03:00:40.676748 136726659987200 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/items/array_metadatas/process_0
I0417 03:01:15.842408 136725049374464 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 37.389636s (commit=36.116745s, array_metadata_write=1.272890s)
I0417 03:01:15.843650 136727733729024 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 38.493 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.03571057319641 s) (per-host)
I0417 03:01:15.843772 136727733729024 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.760550s.
I0417 03:01:17.454131 136727733729024 checkpoint.py:228] Read Metadata={'item_handlers': None, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776394836675496890, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} from gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:01:18.197648 136722335655680 checkpoint.py:247] Updated Metadata={'item_handlers': {'items': 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler'}, 'metrics': {}, 'performance_metrics': {}, 'init_timestamp_nsecs': 1776394836675496890, 'commit_timestamp_nsecs': None, 'custom_metadata': {}} to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/_CHECKPOINT_METADATA
I0417 03:01:18.447538 136727733729024 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 03:01:19.007302 136727733729024 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000237s.
I0417 03:01:19.513608 136727733729024 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.934842s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/items
I0417 03:01:19.515287 136727733729024 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0/items
I0417 03:01:20.041191 136727733729024 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0
I0417 03:01:21.878402 136727733729024 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0`.
I0417 03:01:21.879322 136727733729024 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 47.80 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/0
I0417 03:01:21.880962 136727733729024 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 46.797736s.
I0417 03:01:22.882156 136725040981760 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 03:01:22.882280 136725040981760 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 03:01:22.882346 136725040981760 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts...
I0417 03:01:22.884308 136725040981760 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts.
I0417 03:01:23.176756 136856905627456 metric_logger.py:185] completed step: 4, seconds: 0.578, TFLOP/s/device: 94.052, Tokens/s/device: 14176.591, total_weights: 262144, loss: 10.022
I0417 03:01:23.185395 136856905627456 metric_logger.py:185] completed step: 5, seconds: 0.579, TFLOP/s/device: 93.837, Tokens/s/device: 14144.135, total_weights: 262144, loss: 9.820
I0417 03:01:23.761999 136856905627456 metric_logger.py:185] completed step: 6, seconds: 45.999, TFLOP/s/device: 1.182, Tokens/s/device: 178.090, total_weights: 262144, loss: 9.667
I0417 03:01:24.340376 136856905627456 metric_logger.py:185] completed step: 7, seconds: 0.006, TFLOP/s/device: 8694.371, Tokens/s/device: 1310510.318, total_weights: 262144, loss: 9.561
I0417 03:01:24.918627 136856905627456 metric_logger.py:185] completed step: 8, seconds: 0.578, TFLOP/s/device: 94.089, Tokens/s/device: 14182.138, total_weights: 262144, loss: 9.496
I0417 03:01:25.496678 136856905627456 checkpointing.py:794] Waiting for step 9 to finish before checkpoint...
I0417 03:01:25.497410 136856905627456 checkpointing.py:798] Waited 0.0007531642913818359 seconds for step 9 to finish before starting checkpointing.
I0417 03:01:25.500502 136856905627456 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 03:01:25.502212 136856905627456 checkpoint_manager.py:1512] [process=0] Saving checkpoint at step 9
I0417 03:01:25.503679 136856905627456 event_tracking.py:70] [process=0] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9.
I0417 03:01:26.211936 136856905627456 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array.
I0417 03:01:26.212030 136856905627456 replica_slices.py:424] Transferring arrays to host memory with options: use_replica_parallel=True, min_slice_bytes_for_replica_parallel=None, max_replicas_for_replica_parallel=None, enable_pinned_host_transfer=False
I0417 03:01:26.237305 136856905627456 base_pytree_checkpoint_handler.py:154] [process=0][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.026097s
I0417 03:01:26.238138 136856905627456 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/blocking_gbytes_per_sec: 50.111 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.030783414840698242 s) (per-host)
I0417 03:01:26.238220 136856905627456 base_pytree_checkpoint_handler.py:768] [process=0][thread=MainThread] Initiated Pytree async_save. Time taken: 0.030885s (batch_requests_ready=0.002041s, total_serialization_initiated=0.028088s, others=0.000756s)
I0417 03:01:26.238306 136856905627456 composite_checkpoint_handler.py:715] [process=0][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.034753s (all_items=0.000016s, per_item={'items': '0.00001597'}, temp_paths=0.034737)
I0417 03:01:26.239015 136856905627456 event_tracking.py:125] [process=0] [async] Finished blocking save in 0.74 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9.
I0417 03:01:26.239367 136725040981760 async_checkpointer.py:76] [process=0][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-17 03:21:26.239332
I0417 03:01:26.253768 136856905627456 checkpoint_manager.py:1560] [process=0][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize
I0417 03:01:26.254044 136727163287296 async_checkpointer.py:280] [process=0][thread=save_finalize] Waiting for background save thread=async_save.
I0417 03:01:26.254206 136856905627456 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776394885.5004683, 'wait_for_prev_duration_secs': 7.414817810058594e-05, 'time_between_consecutive_saves_sec': 2.6161186695098877, 'checkpointer_blocking_start_time': 1776394885.5022497, 'checkpointer_blocking_duration_secs': 0.737250566482544, 'get_old_steps_start_time': 1776394886.239526, 'get_old_steps_duration_secs': 2.86102294921875e-05, 'checkpoint_manager_blocking_start_time': 1776394885.4976377, 'checkpoint_manager_blocking_duration_secs': 0.7565319538116455}
I0417 03:01:26.254314 136856905627456 checkpointing.py:409] Started an asynchronous checkpoint save for step 9
I0417 03:01:26.254360 136856905627456 checkpoint_manager.py:2020] [process=0][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete.
I0417 03:01:26.444756 136727733729024 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9
I0417 03:01:28.220109 136727700158208 atomicity.py:140] Creating tmp directory gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9/items
I0417 03:01:31.918693 136727188465408 array_metadata_store.py:203] [process=0][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9/items/array_metadatas/process_0
I0417 03:02:11.770213 136727180072704 base_pytree_checkpoint_handler.py:1282] [process=0][thread=write_metadata_after_commits] Commit + Array metadata written. Time taken: 42.407169s (commit=40.741246s, array_metadata_write=1.665923s)
I0417 03:02:11.771416 136725040981760 base_pytree_checkpoint_handler.py:130] [process=0] /jax/orbax/write/gbytes_per_sec: 34.668 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 45.56402015686035 s) (per-host)
I0417 03:02:11.771544 136725040981760 async_checkpointer.py:90] [process=0][thread=async_save] 3 Handler Commit operations completed. Time taken: 45.532068s.
I0417 03:02:13.434562 136725040981760 ocdbt_utils.py:49] Param validation support for Zarr3 will be added later (b/362328389).
I0417 03:02:13.797940 136725040981760 array_metadata_store.py:411] [process=0][thread=async_save] Validated ArrayMetadata from all 8 hosts. Time taken: 0.000240s.
I0417 03:02:14.465031 136725040981760 base_pytree_checkpoint_handler.py:1406] [process=0][thread=async_save] Pytree save finalize (merge_ocdbt + ArrayMetadata validation) completed. Time taken: 1.920192s. use_zarr3=True, enable_post_merge_validation=True, directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9/items
I0417 03:02:14.466733 136725040981760 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9/items
I0417 03:02:15.005025 136725040981760 atomicity.py:666] Finalizing gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9
I0417 03:02:16.478148 136725040981760 atomicity.py:847] [process=0][thread=async_save] Finished saving checkpoint (finalized tmp dir) to `gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9`.
I0417 03:02:16.479000 136725040981760 event_tracking.py:138] [process=0] [async] Finished save (blocking + background) in 50.98 seconds @ gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_post_train_fixes_20260417_023710/linen_xpk_feat_nnx_post_train_fixes_20260417_023710_06_grad_accum/checkpoints/9
I0417 03:02:16.480586 136725040981760 async_checkpointer.py:160] [process=0][thread=async_save] Background save thread done. Time taken: 50.241107s.
I0417 03:02:16.480734 136727163287296 async_checkpointer.py:288] [process=0][thread=save_finalize] Done with waiting for background save thread=async_save.
I0417 03:02:16.480811 136727163287296 async_checkpointer.py:298] [process=0][thread=save_finalize] No errors found in background save thread=async_save.
I0417 03:02:16.480892 136727163287296 checkpoint_manager.py:2137] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts...
I0417 03:02:16.482718 136727163287296 checkpoint_manager.py:2146] [process=0][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts.
I0417 03:02:16.482884 136856905627456 checkpoint_manager.py:2032] [process=0][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9.
I0417 03:02:16.483036 136856905627456 checkpoint_manager.py:2009] [process=0][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning.
I0417 03:02:16.483653 136856905627456 metric_logger.py:185] completed step: 9, seconds: 0.578, TFLOP/s/device: 93.967, Tokens/s/device: 14163.699, total_weights: 262144, loss: 9.457
Per train step:
 Total TFLOPs: 54.35 
 split as 93.93% learnable weight flops and 6.07% attention flops
XPK End: Fri Apr 17 03:02:28 UTC 2026
EXIT_CODE=0