main (574ad3fb9) vs test/pipeline-scan-nnx (574ad3fb9)| Metric | main 574ad3fb9 | test/pipeline-scan-nnx 574ad3fb9 | Diff (test/pipeline-scan-nnx − main) |
|---|---|---|---|
| Parameters | 1.104 billion | 1.104 billion | — |
| Final loss | 8.1790 | 8.1790 | 0 |
| TFLOP/s | 92.236 | 92.161 | -0.075 |
| Tok/s | 13902.7 | 13891.5 | -11.222 |
| Avg s/step | 2.838 | 2.404 | -0.434 |
| 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 18:23:50 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 18:24:14.732938: 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 18:24:14.915250 133589807769408 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 18:24:23.956416 133589807769408 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 18:24:23.958806 133589807769408 distributed.py:172] Connecting to JAX distributed service on mt-07-eval-87mr2-slice-job-0-0.mt-07-eval-87mr2:8482 I0418 18:24:25.385236 133589807769408 max_utils.py:284] Jax distributed system initialized! I0418 18:24:31.398753 133589807769408 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 18:24:31.398857 133589807769408 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 18:24:31.398896 133589807769408 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 18:24:31.398929 133589807769408 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 18:24:32.094447 133589807769408 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 18:24:32.094719 133589807769408 checkpointing.py:677] Setting up checkpoint logger... I0418 18:24:32.094770 133589807769408 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 18:24:32.094815 133589807769408 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 18:24:32.095158 133589807769408 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 0x797f31914f20>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 18:24:35.798151 133589807769408 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 18:24:35.798380 133589807769408 checkpoint_manager.py:708] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796aa87aebd0>}, handler_registry=None I0418 18:24:35.798613 133589807769408 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796aa87aebd0>` 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 18:24:35.798661 133589807769408 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796aa8415f70>` 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 18:24:35.798696 133589807769408 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 0x796aa87aebd0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x796aa87aebd0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796aa8415f70>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x796aa8415f70>}). I0418 18:24:35.799015 133589807769408 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 18:24:35.799099 133589807769408 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x796aa81cbce0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 18:24:37.367124 133589807769408 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints I0418 18:24:37.369367 133589807769408 checkpoint_manager.py:929] [process=7][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_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x796aa84168d0> I0418 18:24:37.369476 133589807769408 checkpointing.py:301] Checkpoint manager created! I0418 18:24:38.300160 133589807769408 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 18:24:38.300270 133589807769408 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:24:38.677578 133589807769408 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:24:38.677665 133589807769408 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:24:38.693747 133589807769408 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:24:38.693804 133589807769408 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:24:38.717780 133589807769408 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 18:24:38.717847 133589807769408 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:24:38.734154 133589807769408 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 18:24:38.734218 133589807769408 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:24:38.750472 133589807769408 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 18:24:38.750531 133589807769408 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:24:38.775177 133589807769408 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 18:24:38.775246 133589807769408 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:24:38.797477 133589807769408 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 18:24:38.797544 133589807769408 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 18:24:39.020805 133589807769408 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 18:24:39.020917 133589807769408 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 18:24:40.462807 133589807769408 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 18:24:40.462936 133589807769408 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 18:24:40.462992 133589807769408 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 18:24:40.463050 133589807769408 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 18:24:40.463173 133589807769408 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 18:24:40.463227 133589807769408 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 18:24:40.463281 133589807769408 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 18:24:40.463334 133589807769408 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 18:24:40.463375 133589807769408 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 18:24:40.463413 133589807769408 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 18:24:40.463461 133589807769408 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 18:24:40.463510 133589807769408 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 18:24:40.958760 133589807769408 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:24:40.958854 133589807769408 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:24:40.974171 133589807769408 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:24:40.974231 133589807769408 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:24:51.910974 133589807769408 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 18:24:51.911866 133589807769408 metric_logger.py:301] number parameters: 1.104 billion I0418 18:25:03.114929 133589807769408 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 18:25:03.285027 133589807769408 checkpointing.py:776] Waited 0.17008233070373535 seconds for step 0 to finish before starting checkpointing. I0418 18:25:03.287874 133589807769408 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:25:03.290056 133589807769408 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0 I0418 18:25:03.291338 133589807769408 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/0. I0418 18:25:04.034472 133589807769408 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 18:25:04.035513 133589807769408 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:25:04.035569 133589807769408 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 18:25:04.312704 133589807769408 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278303s I0418 18:25:04.312887 133589807769408 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.437 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28371286392211914 s) (per-host) I0418 18:25:04.312945 133589807769408 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.283789s (batch_requests_ready=0.002210s, total_serialization_initiated=0.281494s, others=0.000085s) I0418 18:25:04.313039 133589807769408 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.287995s (all_items=0.000018s, per_item={'items': '0.00001788'}, temp_paths=0.287977) I0418 18:25:04.313897 133589807769408 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.02 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/0. I0418 18:25:04.314238 133460248860416 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 18:45:04.314201 I0418 18:25:04.319727 133589807769408 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:25:04.320004 133458606601984 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:25:04.320145 133589807769408 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776536703.2878551, 'wait_for_prev_duration_secs': 6.580352783203125e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776536703.2901142, 'checkpointer_blocking_duration_secs': 1.024277925491333, 'get_old_steps_start_time': 1776536704.3144164, 'get_old_steps_duration_secs': 3.218650817871094e-05, 'checkpoint_manager_blocking_start_time': 1776536703.2855773, 'checkpoint_manager_blocking_duration_secs': 1.034532070159912} I0418 18:25:04.320305 133589807769408 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 18:25:04.320358 133589807769408 max_utils.py:750] Memstats: After params initialized: I0418 18:25:04.320403 133589807769408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0418 18:25:04.320435 133589807769408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0418 18:25:04.320461 133589807769408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0418 18:25:04.320486 133589807769408 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0418 18:25:04.633824 133589807769408 metric_logger.py:196] completed step: 0, seconds: 11.203, TFLOP/s/device: 1.213, Tokens/s/device: 182.809, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:25:04.814725 133589807769408 metric_logger.py:196] completed step: 1, seconds: 1.517, TFLOP/s/device: 8.955, Tokens/s/device: 1349.739, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:25:05.225471 133589807769408 metric_logger.py:196] completed step: 2, seconds: 0.034, TFLOP/s/device: 402.629, Tokens/s/device: 60688.674, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129 I0418 18:25:05.372825 133589807769408 metric_logger.py:196] completed step: 3, seconds: 0.411, TFLOP/s/device: 33.079, Tokens/s/device: 4985.953, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.203 I0418 18:25:05.667898 133589807769408 metric_logger.py:196] completed step: 4, seconds: 0.153, TFLOP/s/device: 88.698, Tokens/s/device: 13369.455, total_weights: 65536, loss: 9.277, lm_loss: 9.277, perplexity: 10694.614 I0418 18:25:05.673548 133589807769408 metric_logger.py:196] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.248, Tokens/s/device: 13904.636, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.717 I0418 18:25:07.298446 2870 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 18:25:09.739980 133459160409856 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/0/items/array_metadatas/process_7 I0418 18:25:28.505972 133589807769408 metric_logger.py:196] completed step: 6, seconds: 0.295, TFLOP/s/device: 46.000, Tokens/s/device: 6933.583, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.759 I0418 18:25:28.653433 133589807769408 metric_logger.py:196] completed step: 7, seconds: 22.686, TFLOP/s/device: 0.599, Tokens/s/device: 90.277, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4376.905 I0418 18:25:28.800796 133589807769408 metric_logger.py:196] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.229, Tokens/s/device: 13449.616, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3846.406 I0418 18:25:28.947374 133589807769408 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 18:25:28.948052 133589807769408 checkpointing.py:776] Waited 0.0006988048553466797 seconds for step 9 to finish before starting checkpointing. I0418 18:25:28.950676 133589807769408 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:25:40.560132 133460248860416 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 43.240 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 36.53090572357178 s) (per-host) I0418 18:25:40.560254 133460248860416 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.245905s. I0418 18:25:51.686790 133460248860416 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 47.372425s. I0418 18:25:51.687085 133458606601984 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:25:51.687204 133458606601984 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:25:51.687258 133458606601984 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:25:51.689155 133458606601984 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 18:25:51.689332 133589807769408 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0418 18:25:51.689467 133589807769408 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.738788 seconds. If this number is high, consider checkpointing less frequently. I0418 18:25:51.691538 133589807769408 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9 I0418 18:25:51.693326 133589807769408 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/9. I0418 18:25:52.421032 133589807769408 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:25:52.421142 133589807769408 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 18:25:52.454211 133589807769408 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034292s I0418 18:25:52.454344 133589807769408 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 40.950 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03766942024230957 s) (per-host) I0418 18:25:52.454392 133589807769408 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.037727s (batch_requests_ready=0.001774s, total_serialization_initiated=0.035890s, others=0.000063s) I0418 18:25:52.454470 133589807769408 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.041939s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.041923) I0418 18:25:52.455184 133589807769408 event_tracking.py:125] [process=7] [async] Finished blocking save in 0.76 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/9. I0418 18:25:52.455483 133458606601984 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 18:45:52.455449 I0418 18:25:52.462996 133589807769408 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:25:52.463320 133459160409856 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:25:52.463481 133589807769408 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776536728.9506454, 'wait_for_prev_duration_secs': 22.73878812789917, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776536751.6915777, 'checkpointer_blocking_duration_secs': 0.7640447616577148, 'get_old_steps_start_time': 1776536752.4556463, 'get_old_steps_duration_secs': 3.0040740966796875e-05, 'checkpoint_manager_blocking_start_time': 1776536728.9483082, 'checkpoint_manager_blocking_duration_secs': 23.51513695716858} I0418 18:25:52.463592 133589807769408 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 18:25:52.463636 133589807769408 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:25:59.197089 133437839103744 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_07_eval/checkpoints/9/items/array_metadatas/process_7 I0418 18:26:34.301812 133458606601984 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.713 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.88509559631348 s) (per-host) I0418 18:26:34.301939 133458606601984 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.846343s. I0418 18:26:42.653760 133458606601984 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 50.198149s. I0418 18:26:42.654100 133459160409856 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:26:42.654222 133459160409856 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:26:42.654272 133459160409856 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:26:42.655897 133459160409856 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 18:26:42.656066 133589807769408 checkpoint_manager.py:2032] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0418 18:26:42.656247 133589807769408 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:26:42.657321 133589807769408 metric_logger.py:196] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.236, Tokens/s/device: 13902.749, 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 18:26:55 UTC 2026 EXIT_CODE=0
XPK Start: Sun Apr 19 09:13:23 UTC 2026 2026-04-19 09:13:27.577305: 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 00:00:1776590007.590110 9 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776590007.593905 9 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776590007.605421 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776590007.605440 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776590007.605443 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776590007.605445 9 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-19 09:13:47.316119: 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 09:13:47.828958 138402990782272 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-19 09:13:56,869:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0419 09:13:56.869264 138402990782272 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-19 09:13:56,871:jax._src.distributed:157: Connecting to JAX distributed service on mt-07-eval-6rl7u-slice-job-0-0.mt-07-eval-6rl7u:8482 I0419 09:13:56.871868 138402990782272 distributed.py:157] Connecting to JAX distributed service on mt-07-eval-6rl7u-slice-job-0-0.mt-07-eval-6rl7u:8482 I0419 09:13:57.618957 138402990782272 max_utils.py:284] Jax distributed system initialized! I0419 09:14:04.868327 138402990782272 max_utils.py:800] System Information: Jax Version: 0.8.1 I0419 09:14:04.868441 138402990782272 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0419 09:14:04.868480 138402990782272 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 09:14:04.868516 138402990782272 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 09:14:05.485513 138402990782272 maxtext_utils.py:1398] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 09:14:05.485831 138402990782272 checkpointing.py:677] Setting up checkpoint logger... I0419 09:14:05.485888 138402990782272 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 09:14:05.485931 138402990782272 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0419 09:14:05.486272 138402990782272 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 0x7ddfe2379100>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 09:14:08.767656 138402990782272 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0419 09:14:08.767900 138402990782272 checkpoint_manager.py:709] [process=7][thread=MainThread] CheckpointManager init: checkpointers=None, item_names=('items',), item_handlers={'items': <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dcb3c231df0>}, handler_registry=None I0419 09:14:08.768135 138402990782272 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dcb3c231df0>` 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 09:14:08.768183 138402990782272 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dcb3c12c050>` 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 09:14:08.768220 138402990782272 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 0x7dcb3c231df0>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7dcb3c231df0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dcb3c12c050>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7dcb3c12c050>}). I0419 09:14:08.770872 138402990782272 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0419 09:14:08.770955 138402990782272 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7ddfbd4bb2e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0419 09:14:09.466298 138402990782272 checkpoint_manager.py:1818] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints I0419 09:14:09.895144 138402990782272 checkpoint_manager.py:929] [process=7][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_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7dcb3c1301a0> I0419 09:14:09.895321 138402990782272 checkpointing.py:301] Checkpoint manager created! I0419 09:14:10.784662 138402990782272 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 09:14:10.784771 138402990782272 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:14:11.155409 138402990782272 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 09:14:11.155508 138402990782272 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:14:11.171465 138402990782272 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 09:14:11.171527 138402990782272 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:14:11.195381 138402990782272 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 09:14:11.195451 138402990782272 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:14:11.211473 138402990782272 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 09:14:11.211544 138402990782272 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:14:11.227517 138402990782272 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 09:14:11.227590 138402990782272 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:14:11.251657 138402990782272 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 09:14:11.251729 138402990782272 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:14:11.271362 138402990782272 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 09:14:11.271428 138402990782272 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 09:14:11.466844 138402990782272 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0419 09:14:11.466960 138402990782272 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 09:14:12.889315 138402990782272 maxtext_utils.py:1501] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0419 09:14:12.889444 138402990782272 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wi_0/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 09:14:12.889498 138402990782272 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wi_1/kernel Shape: float32[2048,16,7168] Logical: PartitionSpec('embed', 'layers', 'mlp') Physical: ('fsdp', None, None) I0419 09:14:12.889575 138402990782272 maxtext_utils.py:1501] params/params/decoder/layers/mlp/wo/kernel Shape: float32[7168,16,2048] Logical: PartitionSpec('mlp', 'layers', 'embed') Physical: (None, None, 'fsdp') I0419 09:14:12.889714 138402990782272 maxtext_utils.py:1501] params/params/decoder/layers/post_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0419 09:14:12.889787 138402990782272 maxtext_utils.py:1501] params/params/decoder/layers/pre_self_attention_layer_norm/scale Shape: float32[2048,16] Logical: PartitionSpec('norm', 'layers') Physical: (None, None) I0419 09:14:12.889859 138402990782272 maxtext_utils.py:1501] 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 09:14:12.889921 138402990782272 maxtext_utils.py:1501] 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 09:14:12.889965 138402990782272 maxtext_utils.py:1501] 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 09:14:12.890004 138402990782272 maxtext_utils.py:1501] 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 09:14:12.890053 138402990782272 maxtext_utils.py:1501] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0419 09:14:12.890100 138402990782272 maxtext_utils.py:1501] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0419 09:14:13.392812 138402990782272 train.py:156] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 09:14:13.392914 138402990782272 train.py:156] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0419 09:14:13.408036 138402990782272 train.py:163] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 09:14:13.408101 138402990782272 train.py:163] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0419 09:14:24.291520 138402990782272 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 09:14:24.292399 138402990782272 metric_logger.py:289] number parameters: 1.104 billion I0419 09:14:35.260902 138402990782272 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0419 09:14:35.413997 138402990782272 checkpointing.py:776] Waited 0.1530771255493164 seconds for step 0 to finish before starting checkpointing. I0419 09:14:35.416682 138402990782272 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 09:14:35.418450 138402990782272 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 0 I0419 09:14:35.421765 138402990782272 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/0. I0419 09:14:36.394450 138402990782272 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 09:14:36.395497 138402990782272 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 09:14:36.395571 138402990782272 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 09:14:36.642581 138402990782272 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.248214s I0419 09:14:36.642757 138402990782272 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 1.653 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.9333150386810303 s) (per-host) I0419 09:14:36.642812 138402990782272 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.933382s (batch_requests_ready=0.681524s, total_serialization_initiated=0.251782s, others=0.000075s) I0419 09:14:36.642904 138402990782272 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.937464s (all_items=0.000021s, per_item={'items': '0.00002050'}, temp_paths=0.937443) I0419 09:14:36.643911 138279416489728 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0419 09:14:36.644089 138402990782272 async_checkpointer.py:561] Finished blocking save. Time taken: 1.225565s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/0. I0419 09:14:36.649189 138402990782272 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 09:14:36.649516 138275121522432 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0419 09:14:36.649698 138402990782272 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776590075.4166617, 'wait_for_prev_duration_secs': 6.985664367675781e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776590075.4184902, 'checkpointer_blocking_duration_secs': 1.2257084846496582, 'get_old_steps_start_time': 1776590076.644221, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776590075.4146187, 'checkpoint_manager_blocking_duration_secs': 1.2350397109985352} I0419 09:14:36.649806 138402990782272 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0419 09:14:36.649858 138402990782272 max_utils.py:750] Memstats: After params initialized: I0419 09:14:36.649908 138402990782272 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_26(process=7,(2,6,0,0)) I0419 09:14:36.649941 138402990782272 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_27(process=7,(3,6,0,0)) I0419 09:14:36.649969 138402990782272 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_30(process=7,(2,7,0,0)) I0419 09:14:36.649995 138402990782272 max_utils.py:756] Using (GB) 0.43 / 31.25 (1.376000%) on TPU_31(process=7,(3,7,0,0)) I0419 09:14:36.905492 138402990782272 metric_logger.py:185] completed step: 0, seconds: 10.968, TFLOP/s/device: 1.239, Tokens/s/device: 186.718, total_weights: 65536, loss: 10.874 I0419 09:14:37.078720 138402990782272 metric_logger.py:185] completed step: 1, seconds: 1.643, TFLOP/s/device: 8.268, Tokens/s/device: 1246.287, total_weights: 65536, loss: 10.874 I0419 09:14:37.492216 138402990782272 metric_logger.py:185] completed step: 2, seconds: 0.026, TFLOP/s/device: 521.399, Tokens/s/device: 78590.890, total_weights: 65536, loss: 10.263 I0419 09:14:37.639653 138402990782272 metric_logger.py:185] completed step: 3, seconds: 0.414, TFLOP/s/device: 32.817, Tokens/s/device: 4946.466, total_weights: 65536, loss: 9.734 I0419 09:14:37.934860 138402990782272 metric_logger.py:185] completed step: 4, seconds: 0.153, TFLOP/s/device: 89.017, Tokens/s/device: 13417.542, total_weights: 65536, loss: 9.278 I0419 09:14:37.940587 138402990782272 metric_logger.py:185] completed step: 5, seconds: 0.147, TFLOP/s/device: 92.216, Tokens/s/device: 13899.824, total_weights: 65536, loss: 8.892 I0419 09:14:56.748069 138402990782272 metric_logger.py:185] completed step: 6, seconds: 0.296, TFLOP/s/device: 45.975, Tokens/s/device: 6929.852, total_weights: 65536, loss: 8.592 I0419 09:14:56.895666 138402990782272 metric_logger.py:185] completed step: 7, seconds: 18.661, TFLOP/s/device: 0.728, Tokens/s/device: 109.749, total_weights: 65536, loss: 8.384 I0419 09:14:57.043114 138402990782272 metric_logger.py:185] completed step: 8, seconds: 0.152, TFLOP/s/device: 89.301, Tokens/s/device: 13460.489, total_weights: 65536, loss: 8.255 I0419 09:14:57.190272 138402990782272 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0419 09:14:57.190958 138402990782272 checkpointing.py:776] Waited 0.0007071495056152344 seconds for step 9 to finish before starting checkpointing. I0419 09:14:57.192897 138402990782272 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 09:14:57.584839 2610 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 09:14:59.916846 138279424882432 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/0/items/array_metadatas/process_7 I0419 09:15:29.986106 138279416489728 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 29.103 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 54.27662134170532 s) (per-host) I0419 09:15:29.986222 138279416489728 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 53.342198s. I0419 09:15:40.388294 138279416489728 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 63.744253s. I0419 09:15:40.388622 138275121522432 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 09:15:40.388750 138275121522432 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0419 09:15:40.388802 138275121522432 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 09:15:40.390743 138275121522432 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 09:15:40.390926 138402990782272 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=0][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=0. W0419 09:15:40.391053 138402990782272 checkpoint_manager.py:1458] Waiting for previous save to complete took 43.198160 seconds. If this number is high, consider checkpointing less frequently. I0419 09:15:40.392959 138402990782272 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 9 I0419 09:15:40.396354 138402990782272 async_checkpointer.py:452] [process=7] Started async saving checkpoint to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/9. I0419 09:15:40.953820 138402990782272 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 09:15:40.953922 138402990782272 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 09:15:40.998505 138402990782272 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.046627s I0419 09:15:40.998683 138402990782272 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.263 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.2930893898010254 s) (per-host) I0419 09:15:40.998733 138402990782272 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.293148s (batch_requests_ready=0.244578s, total_serialization_initiated=0.048502s, others=0.000067s) I0419 09:15:40.998823 138402990782272 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.297250s (all_items=0.000017s, per_item={'items': '0.00001740'}, temp_paths=0.297233) I0419 09:15:40.999678 138276203656960 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0419 09:15:40.999839 138402990782272 async_checkpointer.py:561] Finished blocking save. Time taken: 0.606807s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/9. I0419 09:15:41.441361 138402990782272 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 09:15:41.441754 138275121522432 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0419 09:15:41.441937 138402990782272 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776590097.192869, 'wait_for_prev_duration_secs': 43.19815993309021, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776590140.3929982, 'checkpointer_blocking_duration_secs': 0.6069536209106445, 'get_old_steps_start_time': 1776590140.9999733, 'get_old_steps_duration_secs': 2.5510787963867188e-05, 'checkpoint_manager_blocking_start_time': 1776590097.191181, 'checkpoint_manager_blocking_duration_secs': 44.25071859359741} I0419 09:15:41.442050 138402990782272 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0419 09:15:41.442096 138402990782272 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 09:15:46.839059 138279424882432 array_metadata_store.py:203] [process=7][thread=array_type_handler] Wrote 39 array_metadata.ArrayMetadata to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_07_eval/checkpoints/9/items/array_metadatas/process_7 I0419 09:16:22.051949 138276203656960 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 38.204 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 41.34631299972534 s) (per-host) I0419 09:16:22.052072 138276203656960 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 41.052283s. I0419 09:16:32.738827 138276203656960 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 51.739021s. I0419 09:16:32.739120 138275121522432 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 09:16:32.739243 138275121522432 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0419 09:16:32.739292 138275121522432 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 09:16:32.740484 138275121522432 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 09:16:32.740682 138402990782272 checkpoint_manager.py:2036] [process=7][thread=MainThread][step=9][wait_until_finished] Done waiting for Save Finalize thread (save_finalize) running at step=9. I0419 09:16:32.740829 138402990782272 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 09:16:32.741505 138402990782272 metric_logger.py:185] completed step: 9, seconds: 0.147, TFLOP/s/device: 92.161, Tokens/s/device: 13891.527, 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 09:16:44 UTC 2026 EXIT_CODE=0