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 | 83.533 | 83.049 | -0.484 |
| Tok/s | 12591.1 | 12518.0 | -73.036 |
| Avg s/step | 2.979 | 2.470 | -0.509 |
| Memory % | 0.54 | 0.54 | 0 |
| JAX | 0.9.2 | 0.8.1 | — |
Diff = branch value − main value. Green = branch improved. Red = branch regressed.
XPK Start: Sat Apr 18 18:53:39 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 18:54:03.937008: 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 18:54:04.464145 135353012787008 max_utils.py:273] Attempting to initialize the jax distributed system... I0418 18:54:13.505421 135353012787008 distributed.py:149] Starting JAX distributed service on [::]:8482 I0418 18:54:13.507887 135353012787008 distributed.py:172] Connecting to JAX distributed service on mt-11-optimizer-offload-nxhn4-slice-job-0-0.mt-11-optimizer-offload-nxhn4:8482 I0418 18:54:14.310976 135353012787008 max_utils.py:284] Jax distributed system initialized! I0418 18:54:19.535177 135353012787008 max_utils.py:800] System Information: Jax Version: 0.9.2 I0418 18:54:19.535278 135353012787008 max_utils.py:801] System Information: Jaxlib Version: 0.9.2 I0418 18:54:19.535316 135353012787008 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 18:54:19.535351 135353012787008 train_utils.py:348] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0418 18:54:20.226017 135353012787008 maxtext_utils.py:1551] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0418 18:54:20.226308 135353012787008 checkpointing.py:677] Setting up checkpoint logger... I0418 18:54:20.226362 135353012787008 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0418 18:54:20.226406 135353012787008 pytree_checkpoint_handler.py:592] save_device_host_concurrent_bytes=None I0418 18:54:20.226744 135353012787008 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 0x7b199c650f50>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0418 18:54:24.371104 135353012787008 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0418 18:54:24.371395 135353012787008 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 0x7b0624593d70>}, handler_registry=None I0418 18:54:24.371631 135353012787008 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b0624593d70>` 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 18:54:24.371679 135353012787008 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b06244386e0>` 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 18:54:24.371714 135353012787008 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 0x7b0624593d70>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7b0624593d70>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b06244386e0>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7b06244386e0>}). I0418 18:54:24.372030 135353012787008 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.34 I0418 18:54:24.372120 135353012787008 async_checkpointer.py:192] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7b06241cb4c0> timeout: 1200 secs and primary_host=0 for async checkpoint writes I0418 18:54:25.900908 135353012787008 checkpoint_manager.py:1812] Found 0 checkpoint steps in gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints I0418 18:54:25.916210 135353012787008 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_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7b0624437bc0> I0418 18:54:25.916381 135353012787008 checkpointing.py:301] Checkpoint manager created! I0418 18:54:26.842638 135353012787008 nnx_wrappers.py:437] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0418 18:54:26.842762 135353012787008 nnx_wrappers.py:437] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:54:27.221735 135353012787008 attentions.py:1088] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:54:27.221836 135353012787008 attentions.py:1088] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:54:27.238420 135353012787008 attentions.py:1089] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0418 18:54:27.238485 135353012787008 attentions.py:1089] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0418 18:54:27.262498 135353012787008 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 18:54:27.262570 135353012787008 attentions.py:1154] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:54:27.279203 135353012787008 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 18:54:27.279263 135353012787008 attentions.py:1155] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:54:27.295820 135353012787008 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 18:54:27.295889 135353012787008 attentions.py:1156] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:54:27.321593 135353012787008 attentions.py:1197] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0418 18:54:27.321671 135353012787008 attentions.py:1197] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0418 18:54:27.342949 135353012787008 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0418 18:54:27.343014 135353012787008 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0418 18:54:27.565032 135353012787008 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0418 18:54:27.565150 135353012787008 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0418 18:54:29.195803 135353012787008 maxtext_utils.py:1654] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: P('norm',) Physical: (None,) I0418 18:54:29.195934 135353012787008 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 18:54:29.195987 135353012787008 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 18:54:29.196045 135353012787008 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 18:54:29.196114 135353012787008 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 18:54:29.196158 135353012787008 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 18:54:29.196211 135353012787008 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 18:54:29.196263 135353012787008 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 18:54:29.196302 135353012787008 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 18:54:29.196339 135353012787008 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 18:54:29.196387 135353012787008 maxtext_utils.py:1654] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: P('embed_vocab', 'vocab') Physical: ('fsdp', None) I0418 18:54:29.196433 135353012787008 maxtext_utils.py:1654] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: P('vocab', 'embed_vocab') Physical: (None, 'fsdp') I0418 18:54:29.689708 135353012787008 train.py:155] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:54:29.689804 135353012787008 train.py:155] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:54:29.705343 135353012787008 train.py:162] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0418 18:54:29.705401 135353012787008 train.py:162] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0418 18:54:40.659116 135353012787008 max_utils.py:791] Total memory size: 1.3 GB, Output size: 0.1 GB, Temp size: 1.1 GB, Argument size: 0.1 GB, Host temp size: 0.0 GB. I0418 18:54:40.659898 135353012787008 metric_logger.py:301] number parameters: 1.104 billion I0418 18:54:51.781822 135353012787008 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0418 18:54:52.132681 135353012787008 checkpointing.py:776] Waited 0.35083985328674316 seconds for step 0 to finish before starting checkpointing. I0418 18:54:52.135118 135353012787008 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:54:52.136825 135353012787008 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 0 I0418 18:54:52.138296 135353012787008 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints/0. I0418 18:54:53.326014 135353012787008 signaling_client.py:364] Using JaxDistributedSignalingClient I0418 18:54:53.327021 135353012787008 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:54:53.327093 135353012787008 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 18:54:53.604138 135353012787008 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.278199s I0418 18:54:53.604306 135353012787008 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 5.422 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.28448057174682617 s) (per-host) I0418 18:54:53.604360 135353012787008 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.284545s (batch_requests_ready=0.002196s, total_serialization_initiated=0.282277s, others=0.000071s) I0418 18:54:53.604454 135353012787008 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.288660s (all_items=0.000017s, per_item={'items': '0.00001717'}, temp_paths=0.288643) I0418 18:54:53.605273 135353012787008 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.47 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints/0. I0418 18:54:53.605614 135220695115520 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:14:53.605575 I0418 18:54:53.607595 135353012787008 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:54:53.607829 135220287297280 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:54:53.607950 135353012787008 standard_logger.py:34] {'step': 0, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538492.1350935, 'wait_for_prev_duration_secs': 8.606910705566406e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538492.1368759, 'checkpointer_blocking_duration_secs': 1.4688851833343506, 'get_old_steps_start_time': 1776538493.6057856, 'get_old_steps_duration_secs': 3.600120544433594e-05, 'checkpoint_manager_blocking_start_time': 1776538492.1332226, 'checkpoint_manager_blocking_duration_secs': 1.4746909141540527} I0418 18:54:53.608054 135353012787008 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0418 18:54:53.608115 135353012787008 max_utils.py:750] Memstats: After params initialized: I0418 18:54:53.608164 135353012787008 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_26(process=7,(2,6,0,0)) I0418 18:54:53.608196 135353012787008 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_27(process=7,(3,6,0,0)) I0418 18:54:53.608223 135353012787008 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_30(process=7,(2,7,0,0)) I0418 18:54:53.608247 135353012787008 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_31(process=7,(3,7,0,0)) I0418 18:54:53.917315 135353012787008 metric_logger.py:196] completed step: 0, seconds: 11.122, TFLOP/s/device: 1.222, Tokens/s/device: 184.143, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:54:54.111398 135353012787008 metric_logger.py:196] completed step: 1, seconds: 2.134, TFLOP/s/device: 6.367, Tokens/s/device: 959.714, total_weights: 65536, loss: 10.874, lm_loss: 10.874, perplexity: 52779.875 I0418 18:54:54.518411 135353012787008 metric_logger.py:196] completed step: 2, seconds: 0.032, TFLOP/s/device: 430.941, Tokens/s/device: 64956.072, total_weights: 65536, loss: 10.262, lm_loss: 10.262, perplexity: 28637.129 I0418 18:54:54.681302 135353012787008 metric_logger.py:196] completed step: 3, seconds: 0.407, TFLOP/s/device: 33.391, Tokens/s/device: 5033.005, total_weights: 65536, loss: 9.734, lm_loss: 9.734, perplexity: 16889.047 I0418 18:54:55.007839 135353012787008 metric_logger.py:196] completed step: 4, seconds: 0.169, TFLOP/s/device: 80.542, Tokens/s/device: 12140.109, total_weights: 65536, loss: 9.278, lm_loss: 9.278, perplexity: 10694.968 I0418 18:54:55.013840 135353012787008 metric_logger.py:196] completed step: 5, seconds: 0.163, TFLOP/s/device: 83.418, Tokens/s/device: 12573.597, total_weights: 65536, loss: 8.892, lm_loss: 8.892, perplexity: 7270.520 I0418 18:54:57.486065 2745 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0418 18:54:59.919502 135220295689984 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_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_7 I0418 18:55:18.421239 135353012787008 metric_logger.py:196] completed step: 6, seconds: 0.327, TFLOP/s/device: 41.545, Tokens/s/device: 6262.097, total_weights: 65536, loss: 8.592, lm_loss: 8.592, perplexity: 5390.688 I0418 18:55:18.584174 135353012787008 metric_logger.py:196] completed step: 7, seconds: 23.245, TFLOP/s/device: 0.585, Tokens/s/device: 88.104, total_weights: 65536, loss: 8.384, lm_loss: 8.384, perplexity: 4377.010 I0418 18:55:18.746783 135353012787008 metric_logger.py:196] completed step: 8, seconds: 0.168, TFLOP/s/device: 80.909, Tokens/s/device: 12195.558, total_weights: 65536, loss: 8.255, lm_loss: 8.255, perplexity: 3845.943 I0418 18:55:18.909330 135353012787008 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0418 18:55:18.909998 135353012787008 checkpointing.py:776] Waited 0.0006895065307617188 seconds for step 9 to finish before starting checkpointing. I0418 18:55:18.912069 135353012787008 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:55:30.456211 135220695115520 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 42.533 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 37.13633322715759 s) (per-host) I0418 18:55:30.456341 135220695115520 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 36.850617s. I0418 18:55:41.277040 135220695115520 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 47.671303s. I0418 18:55:41.277358 135220287297280 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:55:41.277483 135220287297280 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:55:41.277534 135220287297280 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:55:41.279431 135220287297280 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0418 18:55:41.279622 135353012787008 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 18:55:41.279758 135353012787008 checkpoint_manager.py:1452] Waiting for previous save to complete took 22.367696 seconds. If this number is high, consider checkpointing less frequently. I0418 18:55:41.281535 135353012787008 checkpoint_manager.py:1512] [process=7] Saving checkpoint at step 9 I0418 18:55:41.283825 135353012787008 event_tracking.py:70] [process=7] [async] Started save checkpoint @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints/9. I0418 18:55:42.432106 135353012787008 jax_array_handlers.py:360] Scheduling D2H of 39 prioritized jax.Array. I0418 18:55:42.432196 135353012787008 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 18:55:42.466065 135353012787008 base_pytree_checkpoint_handler.py:154] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.034958s I0418 18:55:42.466263 135353012787008 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/blocking_gbytes_per_sec: 40.267 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.03830742835998535 s) (per-host) I0418 18:55:42.466330 135353012787008 base_pytree_checkpoint_handler.py:768] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.038386s (batch_requests_ready=0.001776s, total_serialization_initiated=0.036521s, others=0.000089s) I0418 18:55:42.466456 135353012787008 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.042544s (all_items=0.000015s, per_item={'items': '0.00001526'}, temp_paths=0.042529) I0418 18:55:42.467213 135353012787008 event_tracking.py:125] [process=7] [async] Finished blocking save in 1.19 seconds. Continuing save @ gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints/9. I0418 18:55:42.467512 135220287297280 async_checkpointer.py:76] [process=7][thread=async_save] Background save thread started. Deadline for this save operation is 2026-04-18 19:15:42.467479 I0418 18:55:42.469551 135353012787008 checkpoint_manager.py:1560] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0418 18:55:42.469838 135197685167872 async_checkpointer.py:280] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0418 18:55:42.470008 135353012787008 standard_logger.py:34] {'step': 9, 'event_type': 'save', 'directory': 'gs://lance-maxtext/linen_ckpt_xpk_main_20260418_180002/linen_xpk_main_20260418_180002_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776538518.91203, 'wait_for_prev_duration_secs': 22.367696046829224, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776538541.2815745, 'checkpointer_blocking_duration_secs': 1.186082124710083, 'get_old_steps_start_time': 1776538542.467691, 'get_old_steps_duration_secs': 2.956390380859375e-05, 'checkpoint_manager_blocking_start_time': 1776538518.910248, 'checkpoint_manager_blocking_duration_secs': 23.55972671508789} I0418 18:55:42.470137 135353012787008 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0418 18:55:42.470185 135353012787008 checkpoint_manager.py:2020] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0418 18:55:49.308698 135220295689984 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_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_7 I0418 18:56:25.114351 135220287297280 base_pytree_checkpoint_handler.py:130] [process=7] /jax/orbax/write/gbytes_per_sec: 37.003 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 42.686362504959106 s) (per-host) I0418 18:56:25.114476 135220287297280 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 42.646852s. I0418 18:56:34.653611 135220287297280 async_checkpointer.py:160] [process=7][thread=async_save] Background save thread done. Time taken: 52.185973s. I0418 18:56:34.653889 135197685167872 async_checkpointer.py:288] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0418 18:56:34.654017 135197685167872 async_checkpointer.py:298] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0418 18:56:34.654067 135197685167872 checkpoint_manager.py:2137] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0418 18:56:34.655989 135197685167872 checkpoint_manager.py:2146] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0418 18:56:34.656190 135353012787008 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 18:56:34.656336 135353012787008 checkpoint_manager.py:2009] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0418 18:56:34.657257 135353012787008 metric_logger.py:196] completed step: 9, seconds: 0.163, TFLOP/s/device: 83.533, Tokens/s/device: 12591.067, total_weights: 65536, loss: 8.179, lm_loss: 8.179, perplexity: 3564.739 Per train step: Total TFLOPs: 13.59 split as 93.93% learnable weight flops and 6.07% attention flops XPK End: Sat Apr 18 18:56:46 UTC 2026 EXIT_CODE=0
XPK Start: Sun Apr 19 09:37:09 UTC 2026 2026-04-19 09:37:13.360051: 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 00:00:1776591433.372773 10 cuda_dnn.cc:8579] Unable to register cuDNN factory: Attempting to register factory for plugin cuDNN when one has already been registered E0000 00:00:1776591433.376457 10 cuda_blas.cc:1407] Unable to register cuBLAS factory: Attempting to register factory for plugin cuBLAS when one has already been registered W0000 00:00:1776591433.387695 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776591433.387712 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776591433.387714 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. W0000 00:00:1776591433.387716 10 computation_placer.cc:177] computation placer already registered. Please check linkage and avoid linking the same target more than once. 2026-04-19 09:37:32.680432: 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 09:37:33.201935 139042812180288 max_utils.py:273] Attempting to initialize the jax distributed system... INFO:2026-04-19 09:37:42,243:jax._src.distributed:140: Starting JAX distributed service on [::]:8482 I0419 09:37:42.243474 139042812180288 distributed.py:140] Starting JAX distributed service on [::]:8482 INFO:2026-04-19 09:37:42,245:jax._src.distributed:157: Connecting to JAX distributed service on mt-11-optimizer-offload-b2878-slice-job-0-0.mt-11-optimizer-offload-b2878:8482 I0419 09:37:42.245780 139042812180288 distributed.py:157] Connecting to JAX distributed service on mt-11-optimizer-offload-b2878-slice-job-0-0.mt-11-optimizer-offload-b2878:8482 I0419 09:37:43.216028 139042812180288 max_utils.py:284] Jax distributed system initialized! I0419 09:37:50.259234 139042812180288 max_utils.py:800] System Information: Jax Version: 0.8.1 I0419 09:37:50.259341 139042812180288 max_utils.py:801] System Information: Jaxlib Version: 0.8.1 I0419 09:37:50.259385 139042812180288 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 09:37:50.259419 139042812180288 train_utils.py:347] WARNING: Sequence packing is essentially ignored for synthetic data. Please use a real dataset to use sequence packing. I0419 09:37:50.881775 139042812180288 maxtext_utils.py:1398] Num_devices: 32, shape (1, 1, 1, 32, 1, 1, 1, 1, 1, 1, 1, 1, 1) I0419 09:37:50.882064 139042812180288 checkpointing.py:677] Setting up checkpoint logger... I0419 09:37:50.882118 139042812180288 checkpointing.py:233] Creating checkpoint manager with ocdbt=True and zarr3=True I0419 09:37:50.882161 139042812180288 pytree_checkpoint_handler.py:589] save_device_host_concurrent_bytes=None I0419 09:37:50.882523 139042812180288 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 0x7e74ee1750d0>, enable_pinned_host_transfer=False, save_concurrent_bytes: 96000000000 (89.4 GiB), restore_concurrent_bytes: 96000000000 (89.4 GiB) I0419 09:37:53.685403 139042812180288 checkpointing.py:265] Enabling policy for fixed interval checkpointing. I0419 09:37:53.685775 139042812180288 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 0x7e6058703f80>}, handler_registry=None I0419 09:37:53.686013 139042812180288 composite_checkpoint_handler.py:237] Deferred registration for item: "items". Adding handler `<orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6058703f80>` 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 09:37:53.686062 139042812180288 composite_checkpoint_handler.py:237] Deferred registration for item: "metrics". Adding handler `<orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e609826c050>` 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 09:37:53.686097 139042812180288 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 0x7e6058703f80>, ('items', <class 'orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeRestoreArgs'>): <orbax.checkpoint._src.handlers.pytree_checkpoint_handler.PyTreeCheckpointHandler object at 0x7e6058703f80>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonSaveArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e609826c050>, ('metrics', <class 'orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonRestoreArgs'>): <orbax.checkpoint._src.handlers.json_checkpoint_handler.JsonCheckpointHandler object at 0x7e609826c050>}). I0419 09:37:53.688772 139042812180288 abstract_checkpointer.py:35] orbax-checkpoint version: 0.11.33 I0419 09:37:53.688856 139042812180288 async_checkpointer.py:177] [process=7][thread=MainThread] Using barrier_sync_fn: <function get_barrier_sync_fn.<locals>._fn at 0x7e74d9dfb2e0> timeout: 600 secs and primary_host=0 for async checkpoint writes I0419 09:37:54.942952 139042812180288 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_11_optimizer_offload_true/checkpoints I0419 09:37:55.400102 139042812180288 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_11_optimizer_offload_true/checkpoints: <orbax.checkpoint.checkpoint_manager.CheckpointManager object at 0x7e605844a720> I0419 09:37:55.400282 139042812180288 checkpointing.py:301] Checkpoint manager created! I0419 09:37:56.291622 139042812180288 nnx_wrappers.py:453] Unknown Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_norm_length', 'activation_embed'). I0419 09:37:56.291729 139042812180288 nnx_wrappers.py:453] Unknown Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:37:56.660006 139042812180288 attentions.py:1084] attentions/inputs_q Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 09:37:56.660098 139042812180288 attentions.py:1084] attentions/inputs_q Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:37:56.675733 139042812180288 attentions.py:1085] attentions/inputs_kv Logical: bfloat16[32,2048,2048]...................................... ('activation_batch', 'activation_attn_length', 'activation_attn_embed'). I0419 09:37:56.675793 139042812180288 attentions.py:1085] attentions/inputs_kv Physical: bfloat16[32,2048,2048]...................................... ('fsdp', None, None). I0419 09:37:56.699136 139042812180288 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 09:37:56.699206 139042812180288 attentions.py:1150] attentions/query Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:37:56.715058 139042812180288 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 09:37:56.715121 139042812180288 attentions.py:1151] attentions/key Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:37:56.730902 139042812180288 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 09:37:56.730962 139042812180288 attentions.py:1152] attentions/value Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:37:56.754775 139042812180288 attentions.py:1193] attentions/out Logical: bfloat16[32,2048,16,128].................................... ('activation_batch', 'activation_attn_length', 'activation_heads', 'activation_kv'). I0419 09:37:56.754847 139042812180288 attentions.py:1193] attentions/out Physical: bfloat16[32,2048,16,128].................................... ('fsdp', None, None, None). I0419 09:37:56.774258 139042812180288 linears.py:525] linears/x Logical: bfloat16[32,2048,7168]...................................... ('activation_batch', 'activation_length', 'activation_mlp'). I0419 09:37:56.774323 139042812180288 linears.py:525] linears/x Physical: bfloat16[32,2048,7168]...................................... ('fsdp', None, None). I0419 09:37:56.966447 139042812180288 checkpointing.py:577] checkpoint manager exists so trying to load this run's existing checkpoint I0419 09:37:56.966594 139042812180288 checkpointing.py:665] No existing checkpoints found, not restoring checkpoint. fsdp: 32 I0419 09:37:58.795753 139042812180288 maxtext_utils.py:1501] params/params/decoder/decoder_norm/scale Shape: float32[2048] Logical: PartitionSpec('norm',) Physical: (None,) I0419 09:37:58.795890 139042812180288 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 09:37:58.795945 139042812180288 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 09:37:58.796007 139042812180288 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 09:37:58.796061 139042812180288 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 09:37:58.796100 139042812180288 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 09:37:58.796154 139042812180288 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 09:37:58.796208 139042812180288 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 09:37:58.796249 139042812180288 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 09:37:58.796285 139042812180288 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 09:37:58.796332 139042812180288 maxtext_utils.py:1501] params/params/decoder/logits_dense/kernel Shape: float32[2048,32000] Logical: PartitionSpec('embed', 'vocab') Physical: ('fsdp', None) I0419 09:37:58.796387 139042812180288 maxtext_utils.py:1501] params/params/token_embedder/embedding Shape: float32[32000,2048] Logical: PartitionSpec('vocab', 'embed') Physical: (None, 'fsdp') I0419 09:37:59.301060 139042812180288 train.py:156] train/xent Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 09:37:59.301162 139042812180288 train.py:156] train/xent Physical: float32[32,2048]............................................ ('fsdp', None). I0419 09:37:59.316057 139042812180288 train.py:163] train/z_loss Logical: float32[32,2048]............................................ ('activation_embed_and_logits_batch', 'activation_length'). I0419 09:37:59.316118 139042812180288 train.py:163] train/z_loss Physical: float32[32,2048]............................................ ('fsdp', None). I0419 09:38:10.226606 139042812180288 max_utils.py:791] Total memory size: 1.3 GB, Output size: 0.1 GB, Temp size: 1.1 GB, Argument size: 0.1 GB, Host temp size: 0.0 GB. I0419 09:38:10.227572 139042812180288 metric_logger.py:289] number parameters: 1.104 billion I0419 09:38:21.364496 139042812180288 checkpointing.py:772] Waiting for step 0 to finish before checkpoint... I0419 09:38:21.538565 139042812180288 checkpointing.py:776] Waited 0.1740398406982422 seconds for step 0 to finish before starting checkpointing. I0419 09:38:21.540955 139042812180288 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 09:38:21.542591 139042812180288 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 0 I0419 09:38:21.545323 139042812180288 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_11_optimizer_offload_true/checkpoints/0. I0419 09:38:22.556810 139042812180288 signaling_client.py:364] Using JaxDistributedSignalingClient I0419 09:38:22.557787 139042812180288 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 09:38:22.557846 139042812180288 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 09:38:22.807527 139042812180288 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.250798s I0419 09:38:22.807715 139042812180288 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 3.035 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.5082297325134277 s) (per-host) I0419 09:38:22.807769 139042812180288 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.508294s (batch_requests_ready=0.252988s, total_serialization_initiated=0.255229s, others=0.000077s) I0419 09:38:22.807863 139042812180288 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.512381s (all_items=0.000021s, per_item={'items': '0.00002122'}, temp_paths=0.512360) I0419 09:38:22.808843 138920339678976 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0419 09:38:22.809017 139042812180288 async_checkpointer.py:561] Finished blocking save. Time taken: 1.266353s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_true/checkpoints/0. I0419 09:38:23.252566 139042812180288 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=0] Starting CheckpointManager Save Finalize thread=save_finalize I0419 09:38:23.252945 138908810483456 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0419 09:38:23.253112 139042812180288 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_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776591501.5409381, 'wait_for_prev_duration_secs': 6.389617919921875e-05, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776591501.5426302, 'checkpointer_blocking_duration_secs': 1.2664918899536133, 'get_old_steps_start_time': 1776591502.8091445, 'get_old_steps_duration_secs': 2.4080276489257812e-05, 'checkpoint_manager_blocking_start_time': 1776591501.539151, 'checkpoint_manager_blocking_duration_secs': 1.7139203548431396} I0419 09:38:23.253223 139042812180288 checkpointing.py:408] Started an asynchronous checkpoint save for step 0 I0419 09:38:23.253274 139042812180288 max_utils.py:750] Memstats: After params initialized: I0419 09:38:23.253330 139042812180288 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_26(process=7,(2,6,0,0)) I0419 09:38:23.253365 139042812180288 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_27(process=7,(3,6,0,0)) I0419 09:38:23.253393 139042812180288 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_30(process=7,(2,7,0,0)) I0419 09:38:23.253417 139042812180288 max_utils.py:756] Using (GB) 0.17 / 31.25 (0.544000%) on TPU_31(process=7,(3,7,0,0)) I0419 09:38:23.519212 139042812180288 metric_logger.py:185] completed step: 0, seconds: 11.137, TFLOP/s/device: 1.220, Tokens/s/device: 183.895, total_weights: 65536, loss: 10.874 I0419 09:38:23.704196 139042812180288 metric_logger.py:185] completed step: 1, seconds: 2.153, TFLOP/s/device: 6.310, Tokens/s/device: 951.067, total_weights: 65536, loss: 10.874 I0419 09:38:24.108727 139042812180288 metric_logger.py:185] completed step: 2, seconds: 0.022, TFLOP/s/device: 628.132, Tokens/s/device: 94678.933, total_weights: 65536, loss: 10.263 I0419 09:38:24.272265 139042812180288 metric_logger.py:185] completed step: 3, seconds: 0.405, TFLOP/s/device: 33.561, Tokens/s/device: 5058.714, total_weights: 65536, loss: 9.735 I0419 09:38:24.599940 139042812180288 metric_logger.py:185] completed step: 4, seconds: 0.169, TFLOP/s/device: 80.553, Tokens/s/device: 12141.765, total_weights: 65536, loss: 9.278 I0419 09:38:24.605300 139042812180288 metric_logger.py:185] completed step: 5, seconds: 0.163, TFLOP/s/device: 83.112, Tokens/s/device: 12527.603, total_weights: 65536, loss: 8.892 I0419 09:38:43.430182 139042812180288 metric_logger.py:185] completed step: 6, seconds: 0.328, TFLOP/s/device: 41.421, Tokens/s/device: 6243.350, total_weights: 65536, loss: 8.592 I0419 09:38:43.593827 139042812180288 metric_logger.py:185] completed step: 7, seconds: 18.662, TFLOP/s/device: 0.728, Tokens/s/device: 109.743, total_weights: 65536, loss: 8.384 I0419 09:38:43.757671 139042812180288 metric_logger.py:185] completed step: 8, seconds: 0.168, TFLOP/s/device: 80.709, Tokens/s/device: 12165.277, total_weights: 65536, loss: 8.255 I0419 09:38:43.920991 139042812180288 checkpointing.py:772] Waiting for step 9 to finish before checkpoint... I0419 09:38:43.921591 139042812180288 checkpointing.py:776] Waited 0.00061798095703125 seconds for step 9 to finish before starting checkpointing. I0419 09:38:43.923678 139042812180288 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=0][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 09:38:44.495020 2659 google_auth_provider.cc:181] Running on GCE, using service account 562977990677-compute@developer.gserviceaccount.com I0419 09:38:46.397045 138920348071680 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_11_optimizer_offload_true/checkpoints/0/items/array_metadatas/process_7 I0419 09:39:18.665057 138920339678976 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 28.023 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 56.36553335189819 s) (per-host) I0419 09:39:18.665170 138920339678976 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 55.856217s. I0419 09:39:26.320746 138920339678976 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 63.511779s. I0419 09:39:26.321072 138908810483456 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 09:39:26.321194 138908810483456 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0419 09:39:26.321244 138908810483456 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is syncing with other hosts... I0419 09:39:26.324742 138908810483456 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=0] CheckpointManager Save Finalize is done on all hosts. I0419 09:39:26.324921 139042812180288 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 09:39:26.325051 139042812180288 checkpoint_manager.py:1458] Waiting for previous save to complete took 42.401379 seconds. If this number is high, consider checkpointing less frequently. I0419 09:39:26.327064 139042812180288 checkpoint_manager.py:1518] [process=7] Saving checkpoint at step 9 I0419 09:39:26.330506 139042812180288 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_11_optimizer_offload_true/checkpoints/9. I0419 09:39:27.292207 139042812180288 jax_array_handlers.py:358] Scheduling D2H of 39 prioritized jax.Array. I0419 09:39:27.292312 139042812180288 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 09:39:27.337004 139042812180288 base_pytree_checkpoint_handler.py:153] [process=7][thread=MainThread] Initiated "orbax.checkpoint._src.serialization.jax_array_handlers.ArrayHandler".serialize. Time taken: 0.045905s I0419 09:39:27.337168 139042812180288 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/blocking_gbytes_per_sec: 5.071 GiB/s (total gbytes: 1.5 GiB) (time elapsed: 0.30417728424072266 s) (per-host) I0419 09:39:27.337218 139042812180288 base_pytree_checkpoint_handler.py:737] [process=7][thread=MainThread] Initiated Pytree async_save. Time taken: 0.304238s (batch_requests_ready=0.256500s, total_serialization_initiated=0.047670s, others=0.000067s) I0419 09:39:27.337304 139042812180288 composite_checkpoint_handler.py:715] [process=7][thread=MainThread] Initiated CompositeCheckpointHandler.async_save. Time taken: 0.308420s (all_items=0.000016s, per_item={'items': '0.00001574'}, temp_paths=0.308405) I0419 09:39:27.338272 138909644224256 async_checkpointer.py:79] [process=7][thread=async_save] Background save thread started. I0419 09:39:27.338435 139042812180288 async_checkpointer.py:561] Finished blocking save. Time taken: 1.011297s. Continuing background save to gs://lance-maxtext/linen_ckpt_xpk_test_pipeline_scan_nnx_20260419_085454/linen_xpk_test_pipeline_scan_nnx_20260419_085454_11_optimizer_offload_true/checkpoints/9. I0419 09:39:27.772038 139042812180288 checkpoint_manager.py:1566] [process=7][thread=MainThread][step=9] Starting CheckpointManager Save Finalize thread=save_finalize I0419 09:39:27.772419 138908810483456 async_checkpointer.py:265] [process=7][thread=save_finalize] Waiting for background save thread=async_save. I0419 09:39:27.772629 139042812180288 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_11_optimizer_offload_true/checkpoints', 'reached_preemption': False, 'preemption_received_at': None, 'synchronous': False, 'wait_for_prev_start_time': 1776591523.9236488, 'wait_for_prev_duration_secs': 42.4013786315918, 'time_between_consecutive_saves_sec': None, 'checkpointer_blocking_start_time': 1776591566.3271048, 'checkpointer_blocking_duration_secs': 1.011457920074463, 'get_old_steps_start_time': 1776591567.338583, 'get_old_steps_duration_secs': 2.574920654296875e-05, 'checkpoint_manager_blocking_start_time': 1776591523.9218144, 'checkpoint_manager_blocking_duration_secs': 43.85077214241028} I0419 09:39:27.772738 139042812180288 checkpointing.py:408] Started an asynchronous checkpoint save for step 9 I0419 09:39:27.772784 139042812180288 checkpoint_manager.py:2024] [process=7][thread=MainThread][step=9][wait_until_finished] Waiting for Save Finalize thread (save_finalize) to complete. I0419 09:39:32.574057 138920348071680 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_11_optimizer_offload_true/checkpoints/9/items/array_metadatas/process_7 I0419 09:40:07.861261 138909644224256 base_pytree_checkpoint_handler.py:129] [process=7] /jax/checkpoint/write/gbytes_per_sec: 38.687 MiB/s (total gbytes: 1.5 GiB) (time elapsed: 40.828226804733276 s) (per-host) I0419 09:40:07.861384 138909644224256 async_checkpointer.py:90] [process=7][thread=async_save] 3 Handler Commit operations completed. Time taken: 40.522997s. I0419 09:40:23.174754 138909644224256 async_checkpointer.py:144] [process=7][thread=async_save] Background save thread done. Time taken: 55.836353s. I0419 09:40:23.175057 138908810483456 async_checkpointer.py:273] [process=7][thread=save_finalize] Done with waiting for background save thread=async_save. I0419 09:40:23.175178 138908810483456 async_checkpointer.py:283] [process=7][thread=save_finalize] No errors found in background save thread=async_save. I0419 09:40:23.175226 138908810483456 checkpoint_manager.py:2133] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is syncing with other hosts... I0419 09:40:23.176597 138908810483456 checkpoint_manager.py:2142] [process=7][thread=save_finalize][step=9] CheckpointManager Save Finalize is done on all hosts. I0419 09:40:23.176776 139042812180288 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 09:40:23.176916 139042812180288 checkpoint_manager.py:2013] [process=7][thread=MainThread][wait_until_finished] No Save Finalize thread to wait for. Returning. I0419 09:40:23.177562 139042812180288 metric_logger.py:185] completed step: 9, seconds: 0.164, TFLOP/s/device: 83.049, Tokens/s/device: 12518.031, 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 09:40:34 UTC 2026 EXIT_CODE=0