| [2026-01-10 17:09:55,819] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:621358] bf16 support detected, enabling for this configuration. | |
| [2026-01-10 17:09:56,050] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:621358] baseline 0.000GB () | |
| [2026-01-10 17:09:56,051] [INFO] [axolotl.cli.config.load_cfg:263] [PID:621358] config: | |
| { | |
| "activation_offloading": false, | |
| "adapter": "qlora", | |
| "axolotl_config_path": "olmo-stage1.yaml", | |
| "base_model": "allenai/Olmo-3.1-32B-Instruct", | |
| "base_model_config": "allenai/Olmo-3.1-32B-Instruct", | |
| "batch_size": 2, | |
| "bf16": true, | |
| "capabilities": { | |
| "bf16": true, | |
| "compute_capability": "sm_86", | |
| "fp8": false, | |
| "n_gpu": 1, | |
| "n_node": 1 | |
| }, | |
| "context_parallel_size": 1, | |
| "cut_cross_entropy": true, | |
| "dataloader_num_workers": 1, | |
| "dataloader_pin_memory": true, | |
| "dataloader_prefetch_factor": 256, | |
| "dataset_num_proc": 24, | |
| "dataset_prepared_path": "last_run_prepared", | |
| "datasets": [ | |
| { | |
| "message_property_mappings": { | |
| "content": "content", | |
| "role": "role" | |
| }, | |
| "path": "../marvin_philip_k_dick.json", | |
| "trust_remote_code": false, | |
| "type": "completion" | |
| }, | |
| { | |
| "message_property_mappings": { | |
| "content": "content", | |
| "role": "role" | |
| }, | |
| "path": "../marvin_thomas_ligotti.json", | |
| "trust_remote_code": false, | |
| "type": "completion" | |
| }, | |
| { | |
| "message_property_mappings": { | |
| "content": "content", | |
| "role": "role" | |
| }, | |
| "path": "../marvin_greg_egan.json", | |
| "trust_remote_code": false, | |
| "type": "completion" | |
| }, | |
| { | |
| "message_property_mappings": { | |
| "content": "content", | |
| "role": "role" | |
| }, | |
| "path": "../erotica_quality_trimmed.json", | |
| "trust_remote_code": false, | |
| "type": "completion" | |
| } | |
| ], | |
| "ddp": false, | |
| "device": "cuda:0", | |
| "device_map": "auto", | |
| "dion_rank_fraction": 1.0, | |
| "dion_rank_multiple_of": 1, | |
| "env_capabilities": { | |
| "torch_version": "2.9.1" | |
| }, | |
| "eval_batch_size": 1, | |
| "eval_causal_lm_metrics": [ | |
| "sacrebleu", | |
| "comet", | |
| "ter", | |
| "chrf" | |
| ], | |
| "eval_max_new_tokens": 128, | |
| "eval_sample_packing": true, | |
| "eval_steps": 0.2, | |
| "eval_table_size": 0, | |
| "evals_per_epoch": 5, | |
| "experimental_skip_move_to_device": true, | |
| "flash_attention": false, | |
| "fp16": false, | |
| "gc_steps": 10, | |
| "gradient_accumulation_steps": 2, | |
| "gradient_checkpointing": false, | |
| "group_by_length": false, | |
| "include_tkps": true, | |
| "learning_rate": 1e-05, | |
| "liger_glu_activation": true, | |
| "liger_layer_norm": true, | |
| "liger_rms_norm": true, | |
| "liger_rope": true, | |
| "lisa_layers_attribute": "model.layers", | |
| "load_best_model_at_end": false, | |
| "load_in_4bit": false, | |
| "load_in_8bit": false, | |
| "local_rank": 0, | |
| "logging_steps": 1, | |
| "lora_alpha": 8, | |
| "lora_dropout": 0.05, | |
| "lora_r": 32, | |
| "lora_target_linear": true, | |
| "loraplus_lr_embedding": 1e-06, | |
| "lr_scheduler": "cosine", | |
| "max_grad_norm": 1.0, | |
| "mean_resizing_embeddings": false, | |
| "merge_lora": true, | |
| "micro_batch_size": 1, | |
| "model_config_type": "olmo3", | |
| "num_epochs": 1.0, | |
| "optimizer": "adamw_torch_fused", | |
| "otel_metrics_host": "localhost", | |
| "otel_metrics_port": 8000, | |
| "output_dir": "ckpts-olmo-qlora/marvin", | |
| "pad_to_sequence_len": true, | |
| "peft_use_rslora": true, | |
| "plugins": [ | |
| "axolotl.integrations.liger.LigerPlugin", | |
| "axolotl.integrations.cut_cross_entropy.CutCrossEntropyPlugin" | |
| ], | |
| "pretrain_multipack_attn": true, | |
| "profiler_steps_start": 0, | |
| "qlora_sharded_model_loading": false, | |
| "ray_num_workers": 1, | |
| "resources_per_worker": { | |
| "GPU": 1 | |
| }, | |
| "sample_packing": true, | |
| "sample_packing_bin_size": 200, | |
| "sample_packing_group_size": 100000, | |
| "save_only_model": false, | |
| "save_safetensors": true, | |
| "save_total_limit": 2, | |
| "saves_per_epoch": 1, | |
| "seed": 69, | |
| "sequence_len": 16384, | |
| "shuffle_before_merging_datasets": false, | |
| "shuffle_merged_datasets": true, | |
| "skip_prepare_dataset": false, | |
| "streaming_multipack_buffer_size": 10000, | |
| "strict": false, | |
| "tensor_parallel_size": 1, | |
| "tiled_mlp_use_original_mlp": true, | |
| "tokenizer_config": "allenai/Olmo-3.1-32B-Instruct", | |
| "tokenizer_save_jinja_files": true, | |
| "torch_dtype": "torch.bfloat16", | |
| "train_on_inputs": false, | |
| "trl": { | |
| "log_completions": false, | |
| "mask_truncated_completions": false, | |
| "ref_model_mixup_alpha": 0.9, | |
| "ref_model_sync_steps": 64, | |
| "scale_rewards": true, | |
| "sync_ref_model": false, | |
| "use_vllm": false, | |
| "vllm_server_host": "0.0.0.0", | |
| "vllm_server_port": 8000 | |
| }, | |
| "use_otel_metrics": false, | |
| "use_ray": false, | |
| "use_wandb": true, | |
| "val_set_size": 0.025, | |
| "vllm": { | |
| "device": "auto", | |
| "dtype": "auto", | |
| "gpu_memory_utilization": 0.9, | |
| "host": "0.0.0.0", | |
| "port": 8000 | |
| }, | |
| "wandb_name": "marvin-qlora", | |
| "wandb_project": "Olmo3", | |
| "warmup_ratio": 0.025, | |
| "weight_decay": 0.01, | |
| "world_size": 1 | |
| } | |
| [2026-01-10 17:09:56,051] [INFO] [axolotl.cli.utils.load.load_model_and_tokenizer:40] [PID:621358] loading tokenizer... allenai/Olmo-3.1-32B-Instruct | |
| [2026-01-10 17:09:56,690] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:621358] EOS: 100257 / <|endoftext|> | |
| [2026-01-10 17:09:56,690] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:621358] BOS: 100257 / <|endoftext|> | |
| [2026-01-10 17:09:56,690] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:282] [PID:621358] PAD: 100277 / <|pad|> | |
| [2026-01-10 17:09:56,690] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:283] [PID:621358] UNK: 100257 / <|endoftext|> | |
| [2026-01-10 17:09:56,691] [INFO] [axolotl.cli.utils.load.load_model_and_tokenizer:43] [PID:621358] loading model... | |
| [2026-01-10 17:09:56,852] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:87] [PID:621358] Patched Trainer.evaluation_loop with nanmean loss calculation | |
| [2026-01-10 17:09:56,853] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:138] [PID:621358] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation | |
| [2026-01-10 17:09:56,854] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:301] [PID:621358] Applying multipack dataloader patch for sample packing... | |
| [2026-01-10 17:09:56,932] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:98] [PID:621358] Applying LIGER to olmo3 with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': None, 'rms_norm': True, 'swiglu': True} | |
| [2026-01-10 17:09:56,997] [INFO] [axolotl.integrations.cut_cross_entropy.pre_model_load:94] [PID:621358] Applying Cut Cross Entropy to model type: olmo3 | |
| Loading checkpoint shards: 0%| | 0/14 [00:00<?, ?it/s] Loading checkpoint shards: 7%|███▉ | 1/14 [00:00<00:10, 1.28it/s] Loading checkpoint shards: 14%|███████▊ | 2/14 [00:01<00:09, 1.28it/s] Loading checkpoint shards: 21%|███████████▊ | 3/14 [00:02<00:08, 1.29it/s] Loading checkpoint shards: 29%|███████████████▋ | 4/14 [00:03<00:07, 1.28it/s] Loading checkpoint shards: 36%|███████████████████▋ | 5/14 [00:03<00:07, 1.28it/s] Loading checkpoint shards: 43%|███████████████████████▌ | 6/14 [00:04<00:06, 1.24it/s] Loading checkpoint shards: 50%|███████████████████████████▌ | 7/14 [00:05<00:05, 1.21it/s] Loading checkpoint shards: 57%|███████████████████████████████▍ | 8/14 [00:06<00:04, 1.20it/s] Loading checkpoint shards: 64%|███████████████████████████████████▎ | 9/14 [00:07<00:04, 1.18it/s] Loading checkpoint shards: 71%|██████████████████████████████████████▌ | 10/14 [00:07<00:02, 1.35it/s] Loading checkpoint shards: 100%|██████████████████████████████████████████████████████| 14/14 [00:07<00:00, 1.78it/s] | |
| [2026-01-10 17:10:05,622] [WARNING] [accelerate.big_modeling.dispatch_model:442] [PID:621358] Some parameters are on the meta device because they were offloaded to the cpu. | |
| Traceback (most recent call last): | |
| File "/home/aibox/axo/bin/axolotl", line 10, in <module> | |
| sys.exit(main()) | |
| ^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/main.py", line 347, in main | |
| cli() | |
| File "/home/aibox/axo/lib/python3.12/site-packages/click/core.py", line 1161, in __call__ | |
| return self.main(*args, **kwargs) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/click/core.py", line 1082, in main | |
| rv = self.invoke(ctx) | |
| ^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/click/core.py", line 1697, in invoke | |
| return _process_result(sub_ctx.command.invoke(sub_ctx)) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/click/core.py", line 1443, in invoke | |
| return ctx.invoke(self.callback, **ctx.params) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/click/core.py", line 788, in invoke | |
| return __callback(*args, **kwargs) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/utils/args.py", line 47, in wrapper | |
| return func(*args, **filtered_kwargs) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/main.py", line 293, in merge_lora | |
| do_cli(config=config, **kwargs) | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/merge_lora.py", line 91, in do_cli | |
| do_merge_lora(cfg=parsed_cfg) | |
| File "/home/aibox/training/axolotl/src/axolotl/telemetry/errors.py", line 127, in wrapper | |
| return func(*args, **kwargs) | |
| ^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/merge_lora.py", line 26, in do_merge_lora | |
| model, tokenizer, processor = load_model_and_tokenizer(cfg=cfg) | |
| ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/cli/utils/load.py", line 45, in load_model_and_tokenizer | |
| model, _ = model_loader.load() | |
| ^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/telemetry/errors.py", line 127, in wrapper | |
| return func(*args, **kwargs) | |
| ^^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/training/axolotl/src/axolotl/loaders/model.py", line 180, in load | |
| self._apply_post_model_load_setup() | |
| File "/home/aibox/training/axolotl/src/axolotl/loaders/model.py", line 232, in _apply_post_model_load_setup | |
| self._configure_embedding_dtypes() | |
| File "/home/aibox/training/axolotl/src/axolotl/loaders/model.py", line 308, in _configure_embedding_dtypes | |
| self._convert_embedding_modules_dtype( | |
| File "/home/aibox/training/axolotl/src/axolotl/loaders/model.py", line 877, in _convert_embedding_modules_dtype | |
| module.to(**dest) | |
| File "/home/aibox/axo/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1371, in to | |
| return self._apply(convert) | |
| ^^^^^^^^^^^^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/torch/nn/modules/module.py", line 957, in _apply | |
| param_applied = fn(param) | |
| ^^^^^^^^^ | |
| File "/home/aibox/axo/lib/python3.12/site-packages/torch/nn/modules/module.py", line 1357, in convert | |
| return t.to( | |
| ^^^^^ | |
| torch.OutOfMemoryError: CUDA out of memory. Tried to allocate 1.91 GiB. GPU 0 has a total capacity of 23.57 GiB of which 1.46 GiB is free. Including non-PyTorch memory, this process has 22.10 GiB memory in use. Of the allocated memory 21.85 GiB is allocated by PyTorch, and 10.00 MiB is reserved by PyTorch but unallocated. If reserved but unallocated memory is large try setting PYTORCH_CUDA_ALLOC_CONF=expandable_segments:True to avoid fragmentation. See documentation for Memory Management (https://pytorch.org/docs/stable/notes/cuda.html#environment-variables) | |