main (574ad3fb9) vs feat/nnx-set-defaults-true (574ad3fb9)| Metric | main 574ad3fb9 | feat/nnx-set-defaults-true 574ad3fb9 | Diff (feat/nnx-set-defaults-true − main) |
|---|---|---|---|
| Parameters | 1.104 billion | 1.104 billion | — |
| Final loss | 8.1790 | 8.1790 | 0 |
| TFLOP/s | 92.236 | 92.106 | -0.13 |
| Tok/s | 13902.8 | 13883.2 | -19.603 |
| Avg s/step | 2.899 | 2.793 | -0.106 |
| Memory % | 1.38 | 1.38 | 0 |
| JAX | 0.9.2 | 0.8.1 | — |
Diff = branch value − main value. Green = branch improved. Red = branch regressed.
XPK Start: Sat Apr 18 18:45:37 UTC 2026 PyTorch was not found. Models won't be available and only tokenizers, configuration and file/data utilities can be used. 2026-04-18 18:46:01.812180: 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) I0418 18:46:02.342470 139355902109504 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 18:46:11.384293 139355902109504 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 18:46:11.386642 139355902109504 distributed.py:172] Connecting to JAX distributed service on mt-10-shardy-false-kxha8-slice-job-0-0.mt-10-shardy-false-kxha8:8482 I0418 18:46:12.177243 139355902109504 max_utils.py:284] Jax distributed system initialized! I0418 18:46:18.292302 139355902109504 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 18:46:18.292402 139355902109504 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 18:46:18.292442 139355902109504 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 I0418 18:46:18.292492 139355902109504 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 18:46:18.996400 139355902109504 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 18:46:18.996696 139355902109504 checkpointing.py:677] Setting up checkpoint logger... I0418 18:46:18.996758 139355902109504 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 18:46:18.996803 139355902109504 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 18:46:18.997133 139355902109504 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 0x7ebdb7b16fc0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 18:46:22.649494 139355902109504 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 18:46:22.649749 139355902109504 checkpoint_manager.py:708] [process=5][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ebc8fcee6f0>}, handler_registry=None I0418 18:46:22.650003 139355902109504 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ebc8fcee6f0>` 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`. I0418 18:46:22.650065 139355902109504 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea94c1768d0>` 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`. I0418 18:46:22.650120 139355902109504 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 0x7ebc8fcee6f0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ebc8fcee6f0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea94c1768d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ea94c1768d0>}). I0418 18:46:22.650441 139355902109504 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 18:46:22.650527 139355902109504 async_checkpointer.py:192] [process=5][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ea85c403880> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 18:46:23.896248 139355902109504 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints I0418 18:46:24.319441 139355902109504 checkpoint_manager.py:929] [process=5][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ea94c172060> I0418 18:46:24.319652 139355902109504 checkpointing.py:301] Checkpoint manager created! I0418 18:46:25.233659 139355902109504 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 18:46:25.233771 139355902109504 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:46:25.617884 139355902109504 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:46:25.617974 139355902109504 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:46:25.634266 139355902109504 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:46:25.634326 139355902109504 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:46:25.657938 139355902109504 attentions.py:1154] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 18:46:25.658007 139355902109504 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:46:25.679402 139355902109504 attentions.py:1155] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 18:46:25.679508 139355902109504 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:46:25.695803 139355902109504 attentions.py:1156] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0418 18:46:25.695863 139355902109504 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:46:25.721276 139355902109504 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 18:46:25.721347 139355902109504 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:46:25.742352 139355902109504 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 18:46:25.742417 139355902109504 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 18:46:25.957729 139355902109504 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 18:46:25.957834 139355902109504 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 18:46:27.341172 139355902109504 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 18:46:27.341293 139355902109504 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 18:46:27.341345 139355902109504 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: P('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0418 18:46:27.341400 139355902109504 maxtext_utils.py:1654] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: P('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0418 18:46:27.341485 139355902109504 maxtext_utils.py:1654] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 18:46:27.341555 139355902109504 maxtext_utils.py:1654] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: P('norm', 'layers') Physical: (None, None) I0418 18:46:27.341618 139355902109504 maxtext_utils.py:1654] 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) I0418 18:46:27.341670 139355902109504 maxtext_utils.py:1654] 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') I0418 18:46:27.341709 139355902109504 maxtext_utils.py:1654] 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) I0418 18:46:27.341745 139355902109504 maxtext_utils.py:1654] 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) I0418 18:46:27.341791 139355902109504 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 18:46:27.341839 139355902109504 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 18:46:27.832659 139355902109504 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:46:27.832752 139355902109504 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:46:27.847918 139355902109504 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:46:27.847978 139355902109504 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:46:38.751034 139355902109504 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0418 18:46:38.751928 139355902109504 metric_logger.py:301] number parameters: 1.104 billion I0418 18:46:50.205392 139355902109504 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 18:46:50.372630 139355902109504 checkpointing.py:776] Waited 0.16721844673156738 seconds for step 0 to finish before starting checkpointing. I0418 18:46:50.375202 139355902109504 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:46:50.377058 139355902109504 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 0 I0418 18:46:50.378554 139355902109504 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/0. I0418 18:46:51.522650 139355902109504 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 18:46:51.523610 139355902109504 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:46:51.523671 139355902109504 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 I0418 18:46:51.794860 139355902109504 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.272285s I0418 18:46:51.795035 139355902109504 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 5.553 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2778048515319824 s) (per-host) I0418 18:46:51.795089 139355902109504 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.277869s (batch_requests_ready=0.002237s, total_serialization_initiated=0.275561s, others=0.000072s) I0418 18:46:51.795179 139355902109504 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.281934s (all_items=0.000017s, per_item={'items': '0.00001669'}, temp_paths=0.281917) I0418 18:46:51.795941 139355902109504 event_tracking.py:125] [process=5] [async] Finished blocking save in 1.42 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/0. I0418 18:46:51.796278 139226800715520 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:06:51.796241 I0418 18:46:51.806780 139355902109504 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:46:51.807093 139226746173184 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:46:51.807246 139355902109504 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538010.3751845, 'wait_for_prev_duration_secs': 6.246566772460938e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538010.3770974, 'checkpointer_blocking_duration_secs': 1.4193315505981445, 'get_old_steps_start_time': 1776538011.7964513, 'get_old_steps_duration_secs': 3.814697265625e-05, 'checkpoint_manager_blocking_start_time': 1776538010.3731267, 'checkpoint_manager_blocking_duration_secs': 1.4340808391571045} I0418 18:46:51.807355 139355902109504 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 18:46:51.807404 139355902109504 max_utils.py:750] Memstats: After params initialized: I0418 18:46:51.807464 139355902109504 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_18(process=5,(2,4,0,0)) I0418 18:46:51.807498 139355902109504 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_19(process=5,(3,4,0,0)) I0418 18:46:51.807527 139355902109504 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_22(process=5,(2,5,0,0)) I0418 18:46:51.807552 139355902109504 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_23(process=5,(3,5,0,0)) I0418 18:46:52.118390 139355902109504 metric_logger.py:196] completed step: 0, seconds: 11.453, TFLOP/s/device: 1.186, Tokens/s/device: 178.812, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:46:52.309910 139355902109504 metric_logger.py:196] completed step: 1, seconds: 1.911, TFLOP/s/device: 7.108, Tokens/s/device: 1071.432, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:46:52.721509 139355902109504 metric_logger.py:196] completed step: 2, seconds: 0.044, TFLOP/s/device: 307.979, Tokens/s/device: 46422.014, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129 I0418 18:46:52.868733 139355902109504 metric_logger.py:196] completed step: 3, seconds: 0.412, TFLOP/s/device: 32.979, Tokens/s/device: 4970.982, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203 I0418 18:46:53.163907 139355902109504 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.940, Tokens/s/device: 13405.949, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614 I0418 18:46:53.169483 139355902109504 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.246, Tokens/s/device: 13904.353, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717 I0418 18:46:55.968588 2809 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 18:46:58.472544 139226754565888 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/0/items/array_metadatas/process_5 I0418 18:47:16.146224 139355902109504 metric_logger.py:196] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.976, Tokens/s/device: 6930.016, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759 I0418 18:47:16.293702 139355902109504 metric_logger.py:196] completed step: 7, seconds: 22.830, TFLOP/s/device: 0.595, Tokens/s/device: 89.707, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905 I0418 18:47:16.440944 139355902109504 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.235, Tokens/s/device: 13450.500, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406 I0418 18:47:16.587806 139355902109504 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 18:47:16.588476 139355902109504 checkpointing.py:776] Waited 0.0006716251373291016 seconds for step 9 to finish before starting checkpointing. I0418 18:47:16.590465 139355902109504 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:47:30.791839 139226800715520 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 40.219 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 39.274569511413574 s) (per-host) I0418 18:47:30.791959 139226800715520 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 38.995572s. I0418 18:47:39.321121 139226800715520 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 47.524718s. I0418 18:47:39.321437 139226746173184 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:47:39.321567 139226746173184 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:47:39.321617 139226746173184 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:47:39.324641 139226746173184 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 18:47:39.324816 139355902109504 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 18:47:39.324952 139355902109504 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.734497 seconds. If this number is high, consider checkpointing less frequently. I0418 18:47:39.326681 139355902109504 checkpoint_manager.py:1512] [process=5] Saving checkpoint at step 9 I0418 18:47:39.328622 139355902109504 event_tracking.py:70] [process=5] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/9. I0418 18:47:40.039915 139355902109504 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:47:40.040009 139355902109504 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 I0418 18:47:40.074328 139355902109504 base_pytree_checkpoint_handler.py:154] [process=5][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.035544s I0418 18:47:40.074491 139355902109504 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/blocking_gbytes_per_sec: 39.529 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03902435302734375 s) (per-host) I0418 18:47:40.074541 139355902109504 base_pytree_checkpoint_handler.py:768] [process=5][thread=MainThread] Initiated Pytree async_save. Time taken: 0.039084s (batch_requests_ready=0.001831s, total_serialization_initiated=0.037186s, others=0.000067s) I0418 18:47:40.074626 139355902109504 composite_checkpoint_handler.py:715] [process=5][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.043243s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.043227) I0418 18:47:40.075281 139355902109504 event_tracking.py:125] [process=5] [async] Finished blocking save in 0.75 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/9. I0418 18:47:40.075621 139226746173184 async_checkpointer.py:76] [process=5][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:07:40.075583 I0418 18:47:40.077700 139355902109504 checkpoint_manager.py:1560] [process=5][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:47:40.078047 139220445353728 async_checkpointer.py:280] [process=5][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:47:40.078239 139355902109504 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538036.590424, 'wait_for_prev_duration_secs': 22.73449683189392, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538059.3267257, 'checkpointer_blocking_duration_secs': 0.7490432262420654, 'get_old_steps_start_time': 1776538060.0757914, 'get_old_steps_duration_secs': 3.123283386230469e-05, 'checkpoint_manager_blocking_start_time': 1776538036.588701, 'checkpoint_manager_blocking_duration_secs': 23.489494800567627} I0418 18:47:40.078377 139355902109504 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 18:47:40.078428 139355902109504 checkpoint_manager.py:2020] [process=5][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:47:46.821598 139226754565888 array_metadata_store.py:203] [process=5][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_10_shardy_false/checkpoints/9/items/array_metadatas/process_5 I0418 18:48:22.178649 139226746173184 base_pytree_checkpoint_handler.py:130] [process=5] /jax/orbax/write/gbytes_per_sec: 37.482 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.14315176010132 s) (per-host) I0418 18:48:22.178758 139226746173184 async_checkpointer.py:90] [process=5][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.103024s. I0418 18:48:32.168947 139226746173184 async_checkpointer.py:160] [process=5][thread=async_save] Background save thread done. Time taken: 52.093197s. I0418 18:48:32.169241 139220445353728 async_checkpointer.py:288] [process=5][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:48:32.169364 139220445353728 async_checkpointer.py:298] [process=5][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:48:32.169418 139220445353728 checkpoint_manager.py:2137] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:48:32.171113 139220445353728 checkpoint_manager.py:2146] [process=5][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 18:48:32.171308 139355902109504 checkpoint_manager.py:2032] [process=5][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 18:48:32.171464 139355902109504 checkpoint_manager.py:2009] [process=5][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:48:32.172433 139355902109504 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.236, Tokens/s/device: 13902.843, total_weights: 65536, loss: 8.179, lm_loss: 8.179, perplexity: 3564.635 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sat Apr 18 18:48:43 UTC 2026 EXIT_CODE=0
XPK Start: Sun Apr 19 03:53:38 UTC 2026 2026-04-19 03:53:42.491712: 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 00:00:1776570822.504365 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776570822.508023 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776570822.519261 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776570822.519280 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776570822.519283 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776570822.519285 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-19 03:54:01.660698: 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) I0419 03:54:02.187041 134995663525696 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-19 03:54:11,226:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0419 03:54:11.226190 134995663525696 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-19 03:54:11,228:jax._src.distributed:157: Connecting to JAX distributed service on mt-10-shardy-false-aybxq-slice-job-0-0.mt-10-shardy-false-aybxq:8482 I0419 03:54:11.228576 134995663525696 distributed.py:157] Connecting to JAX distributed service on mt-10-shardy-false-aybxq-slice-job-0-0.mt-10-shardy-false-aybxq:8482 I0419 03:54:13.060941 134995663525696 max_utils.py:284] Jax distributed system initialized! I0419 03:54:19.953674 134995663525696 max_utils.py:800] System Information: Jax Version: 0.8.1 I0419 03:54:19.953783 134995663525696 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0419 03:54:19.953823 134995663525696 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 I0419 03:54:19.953855 134995663525696 train_utils.py:377] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 03:54:20.579356 134995663525696 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 03:54:20.579952 134995663525696 maxtext_utils.py:1631] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 03:54:20.580132 134995663525696 checkpointing.py:688] Setting up checkpoint logger... I0419 03:54:20.580183 134995663525696 checkpointing.py:234] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 03:54:20.580226 134995663525696 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0419 03:54:20.580583 134995663525696 base_pytree_checkpoint_handler.py:415] 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 0x7ac6a1b79250>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 03:54:24.634653 134995663525696 checkpointing.py:266] Enabling policy for fixed interval checkpointing. I0419 03:54:24.634890 134995663525696 checkpoint_manager.py:709] [process=6][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab37e3196d0>}, handler_registry=None I0419 03:54:24.635125 134995663525696 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab37e3196d0>` 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`. I0419 03:54:24.635174 134995663525696 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab37e31e810>` 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`. I0419 03:54:24.635212 134995663525696 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 0x7ab37e3196d0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7ab37e3196d0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab37e31e810>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7ab37e31e810>}). I0419 03:54:24.635719 134995663525696 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0419 03:54:24.635797 134995663525696 async_checkpointer.py:177] [process=6][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ab33c3a8180> timeout: 600 secs and primary_host=0 for async checkpoint writes I0419 03:54:27.094284 134995663525696 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints I0419 03:54:27.098503 134995663525696 checkpoint_manager.py:929] [process=6][thread=MainThread] CheckpointManager created, primary_host=0, CheckpointManagerOptions=CheckpointManagerOptions(save_interval_steps=1, max_to_keep=None, keep_time_interval=None, keep_period=None, should_keep_fn=None, best_fn=None, best_mode='max', keep_checkpoints_without_metrics=True, step_prefix=None, step_format_fixed_length=None, step_name_format=None, create=True, cleanup_tmp_directories=False, save_on_steps=frozenset(), single_host_load_and_broadcast=False, todelete_subdir=None, todelete_full_path=None, enable_background_delete=False, read_only=False, enable_async_checkpointing=True, async_options=None, multiprocessing_options=MultiprocessingOptions(primary_host=0, active_processes=None, barrier_sync_key_prefix=None), should_save_fn=None, file_options=FileOptions(path_permission_mode=None), save_root_metadata=True, temporary_path_class=None, save_decision_policy=FixedIntervalPolicy(interval=10), preservation_policy=LatestN(n=None), prevent_write_metrics=False, enable_should_save_is_saving_in_progress_check=True, enable_per_process_directory_creation=False, lightweight_initialize=False), root_directory=gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7ab37e31a9f0> I0419 03:54:27.098621 134995663525696 checkpointing.py:302] Checkpoint manager created! I0419 03:54:27.993580 134995663525696 nnx_wrappers.py:455] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 03:54:27.993685 134995663525696 nnx_wrappers.py:455] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 03:54:28.366987 134995663525696 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 03:54:28.367088 134995663525696 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 03:54:28.383156 134995663525696 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 03:54:28.383216 134995663525696 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 03:54:28.405959 134995663525696 attentions.py:1150] attentions/query Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 03:54:28.406027 134995663525696 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 03:54:28.421770 134995663525696 attentions.py:1151] attentions/key Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 03:54:28.421834 134995663525696 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 03:54:28.437510 134995663525696 attentions.py:1152] attentions/value Logical: bfloat16[32,2048,16,128].................................... ('activation_kv_batch', 'activation_attn_length', 'activation_kv_heads', 'activation_kv_head_dim'). I0419 03:54:28.437576 134995663525696 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 03:54:28.462278 134995663525696 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 03:54:28.462347 134995663525696 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 03:54:28.481674 134995663525696 linears.py:541] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 03:54:28.481739 134995663525696 linears.py:541] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 03:54:28.670149 134995663525696 checkpointing.py:578] checkpoint manager exists so trying to load this run's existing checkpoint I0419 03:54:28.670272 134995663525696 checkpointing.py:676] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 03:54:30.091887 134995663525696 maxtext_utils.py:1740] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0419 03:54:30.092022 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 03:54:30.092082 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 03:54:30.092141 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0419 03:54:30.092194 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0419 03:54:30.092247 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0419 03:54:30.092305 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/key/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0419 03:54:30.092363 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/out/kernel Shape: float32[16,16,128,2048] Logical: PartitionSpec('heads', 'layers', 'kv', 'embed') Physical: (None, None, None, 'fsdp') I0419 03:54:30.092405 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/query/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'q_heads', 'kv') Physical: ('fsdp', None, None, None) I0419 03:54:30.092441 134995663525696 maxtext_utils.py:1740] params/params/decoder/layers/self_attention/value/kernel Shape: float32[2048,16,16,128] Logical: PartitionSpec('embed', 'layers', 'kv_heads', 'kv_head_dim') Physical: ('fsdp', None, None, None) I0419 03:54:30.092489 134995663525696 maxtext_utils.py:1740] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0419 03:54:30.092539 134995663525696 maxtext_utils.py:1740] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0419 03:54:30.592093 134995663525696 train.py:158] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 03:54:30.592187 134995663525696 train.py:158] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0419 03:54:30.607172 134995663525696 train.py:165] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 03:54:30.607247 134995663525696 train.py:165] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0419 03:54:41.246995 134995663525696 max_utils.py:791] Total memory size: 1.5 GB, Output size: 0.4 GB, Temp size: 1.1 GB, Argument size: 0.4 GB, Host temp size: 0.0 GB. I0419 03:54:41.247856 134995663525696 metric_logger.py:289] number parameters: 1.104 billion I0419 03:54:52.350228 134995663525696 checkpointing.py:794] Waiting for step 0 to finish before checkpoint... I0419 03:54:52.507860 134995663525696 checkpointing.py:798] Waited 0.15761375427246094 seconds for step 0 to finish before starting checkpointing. I0419 03:54:52.510411 134995663525696 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 03:54:52.512080 134995663525696 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 0 I0419 03:54:52.514806 134995663525696 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/0. I0419 03:54:53.506943 134995663525696 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 03:54:53.508136 134995663525696 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 03:54:53.508195 134995663525696 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 I0419 03:54:53.758790 134995663525696 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.251930s I0419 03:54:53.758960 134995663525696 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.042 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5070748329162598 s) (per-host) I0419 03:54:53.759011 134995663525696 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.507134s (batch_requests_ready=0.251547s, total_serialization_initiated=0.255520s, others=0.000067s) I0419 03:54:53.759100 134995663525696 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.511336s (all_items=0.000019s, per_item={'items': '0.00001931'}, temp_paths=0.511317) I0419 03:54:53.760034 134872702117632 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0419 03:54:53.760204 134995663525696 async_checkpointer.py:561] Finished blocking save. Time taken: 1.248053s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/0. I0419 03:54:54.618680 134995663525696 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 03:54:54.619070 134867878672128 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0419 03:54:54.619251 134995663525696 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776570892.5103946, 'wait_for_prev_duration_secs': 6.4849853515625e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776570892.512119, 'checkpointer_blocking_duration_secs': 1.2482068538665771, 'get_old_steps_start_time': 1776570893.760348, 'get_old_steps_duration_secs': 2.5272369384765625e-05, 'checkpoint_manager_blocking_start_time': 1776570892.5084581, 'checkpoint_manager_blocking_duration_secs': 2.11073899269104} I0419 03:54:54.619359 134995663525696 checkpointing.py:409] Started an asynchronous checkpoint save for step 0 I0419 03:54:54.619411 134995663525696 max_utils.py:750] Memstats: After params initialized: I0419 03:54:54.619467 134995663525696 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_24(process=6,(0,6,0,0)) I0419 03:54:54.619503 134995663525696 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_25(process=6,(1,6,0,0)) I0419 03:54:54.619536 134995663525696 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_28(process=6,(0,7,0,0)) I0419 03:54:54.619563 134995663525696 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_29(process=6,(1,7,0,0)) I0419 03:54:54.882458 134995663525696 metric_logger.py:185] completed step: 0, seconds: 11.102, TFLOP/s/device: 1.224, Tokens/s/device: 184.467, total_weights: 65536, loss: 10.874 I0419 03:54:55.063828 134995663525696 metric_logger.py:185] completed step: 1, seconds: 2.531, TFLOP/s/device: 5.368, Tokens/s/device: 809.196, total_weights: 65536, loss: 10.874 I0419 03:54:55.859093 134995663525696 metric_logger.py:185] completed step: 2, seconds: 0.034, TFLOP/s/device: 397.087, Tokens/s/device: 59853.289, total_weights: 65536, loss: 10.263 I0419 03:54:56.006720 134995663525696 metric_logger.py:185] completed step: 3, seconds: 0.796, TFLOP/s/device: 17.074, Tokens/s/device: 2573.550, total_weights: 65536, loss: 9.734 I0419 03:54:56.302023 134995663525696 metric_logger.py:185] completed step: 4, seconds: 0.153, TFLOP/s/device: 89.035, Tokens/s/device: 13420.356, total_weights: 65536, loss: 9.278 I0419 03:54:56.307708 134995663525696 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.214, Tokens/s/device: 13899.446, total_weights: 65536, loss: 8.892 I0419 03:55:17.329708 134995663525696 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.930, Tokens/s/device: 6923.035, total_weights: 65536, loss: 8.592 I0419 03:55:17.477285 134995663525696 metric_logger.py:185] completed step: 7, seconds: 20.876, TFLOP/s/device: 0.651, Tokens/s/device: 98.105, total_weights: 65536, loss: 8.384 I0419 03:55:17.624789 134995663525696 metric_logger.py:185] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.539, Tokens/s/device: 13496.237, total_weights: 65536, loss: 8.255 I0419 03:55:17.771954 134995663525696 checkpointing.py:794] Waiting for step 9 to finish before checkpoint... I0419 03:55:17.772646 134995663525696 checkpointing.py:798] Waited 0.0007131099700927734 seconds for step 9 to finish before starting checkpointing. I0419 03:55:17.774527 134995663525696 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 03:55:18.590979 2622 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 03:55:21.177212 134872710510336 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/0/items/array_metadatas/process_6 I0419 03:55:50.934852 134872702117632 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 27.384 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 57.682924032211304 s) (per-host) I0419 03:55:50.934976 134872702117632 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 57.174831s. I0419 03:56:01.503636 134872702117632 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 67.743475s. I0419 03:56:01.503868 134867878672128 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 03:56:01.503929 134867878672128 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0419 03:56:01.503981 134867878672128 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 03:56:01.505697 134867878672128 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 03:56:01.505865 134995663525696 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 03:56:01.506011 134995663525696 checkpoint_manager.py:1458] Waiting for previous save to complete took 43.731487 seconds. If this number is high, consider checkpointing less frequently. I0419 03:56:01.507746 134995663525696 checkpoint_manager.py:1518] [process=6] Saving checkpoint at step 9 I0419 03:56:01.511144 134995663525696 async_checkpointer.py:452] [process=6] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/9. I0419 03:56:03.341085 134995663525696 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 03:56:03.341182 134995663525696 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 I0419 03:56:03.390259 134995663525696 base_pytree_checkpoint_handler.py:153] [process=6][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.050360s I0419 03:56:03.390451 134995663525696 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/blocking_gbytes_per_sec: 2.079 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.7420163154602051 s) (per-host) I0419 03:56:03.390524 134995663525696 base_pytree_checkpoint_handler.py:737] [process=6][thread=MainThread] Initiated Pytree async_save. Time taken: 0.742103s (batch_requests_ready=0.689782s, total_serialization_initiated=0.052223s, others=0.000098s) I0419 03:56:03.390633 134995663525696 composite_checkpoint_handler.py:715] [process=6][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.746057s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.746042) I0419 03:56:03.391520 134869973694208 async_checkpointer.py:79] [process=6][thread=async_save] Background save thread started. I0419 03:56:03.391676 134995663525696 async_checkpointer.py:561] Finished blocking save. Time taken: 1.883858s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/9. I0419 03:56:03.394783 134995663525696 checkpoint_manager.py:1566] [process=6][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 03:56:03.395067 134867878672128 async_checkpointer.py:265] [process=6][thread=save_finalize] Waiting for background save thread=async_save. I0419 03:56:03.395241 134995663525696 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776570917.7745001, 'wait_for_prev_duration_secs': 43.73148703575134, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776570961.507786, 'checkpointer_blocking_duration_secs': 1.8839983940124512, 'get_old_steps_start_time': 1776570963.3918045, 'get_old_steps_duration_secs': 2.6226043701171875e-05, 'checkpoint_manager_blocking_start_time': 1776570917.7728758, 'checkpoint_manager_blocking_duration_secs': 45.62231755256653} I0419 03:56:03.395343 134995663525696 checkpointing.py:409] Started an asynchronous checkpoint save for step 9 I0419 03:56:03.395387 134995663525696 checkpoint_manager.py:2024] [process=6][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 03:56:09.716909 134872710510336 array_metadata_store.py:203] [process=6][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_feat_nnx_set_defaults_true_20260419_025828/linen_xpk_feat_nnx_set_defaults_true_20260419_025828_10_shardy_false/checkpoints/9/items/array_metadatas/process_6 I0419 03:56:46.435780 134869973694208 base_pytree_checkpoint_handler.py:129] [process=6] /jax/checkpoint/write/gbytes_per_sec: 36.074 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 43.787309885025024 s) (per-host) I0419 03:56:46.435896 134869973694208 async_checkpointer.py:90] [process=6][thread=async_save] 3 Handler Commit operations completed. Time taken: 43.044266s. I0419 03:56:53.950583 134869973694208 async_checkpointer.py:144] [process=6][thread=async_save] Background save thread done. Time taken: 50.558937s. I0419 03:56:53.950870 134867878672128 async_checkpointer.py:273] [process=6][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 03:56:53.950997 134867878672128 async_checkpointer.py:283] [process=6][thread=save_finalize] No errors found in background save thread=async_save. I0419 03:56:53.951042 134867878672128 checkpoint_manager.py:2133] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 03:56:53.952372 134867878672128 checkpoint_manager.py:2142] [process=6][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 03:56:53.952547 134995663525696 checkpoint_manager.py:2036] [process=6][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 03:56:53.952683 134995663525696 checkpoint_manager.py:2013] [process=6][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 03:56:53.953324 134995663525696 metric_logger.py:185] completed step: 9, seconds: 0.148, TFLOP/s/device: 92.106, Tokens/s/device: 13883.240, total_weights: 65536, loss: 8.179 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sun Apr 19 03:57:04 UTC 2026 EXIT_CODE=0