[2026-03-20 00:33:26,556] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:13067] baseline 0.000GB () [2026-03-20 00:33:26,556] [INFO] [axolotl.cli.config.load_cfg:340] [PID:13067] config: { "activation_offloading": false, "adapter": "lora", "axolotl_config_path": "writer.yaml", "base_model": "allenai/Olmo-3.1-32B-Instruct", "base_model_config": "allenai/Olmo-3.1-32B-Instruct", "batch_size": 16, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_90", "fp8": true, "n_gpu": 1, "n_node": 1 }, "chat_template": "chatml", "context_parallel_size": 1, "dataloader_num_workers": 1, "dataloader_pin_memory": true, "dataloader_prefetch_factor": 256, "dataset_num_proc": 28, "datasets": [ { "chat_template": "tokenizer_default", "message_field_training": "train", "message_property_mappings": { "content": "content", "role": "role" }, "path": "ConicCat/C2_Sonnet_4_5", "roles_to_train": [], "trust_remote_code": false, "type": "chat_template" }, { "chat_template": "tokenizer_default", "message_property_mappings": { "content": "content", "role": "role" }, "path": "ConicCat/Gutenberg-SFT", "trust_remote_code": false, "type": "chat_template" }, { "chat_template": "tokenizer_default", "message_property_mappings": { "content": "content", "role": "role" }, "path": "ConicCat/Condor-SFT-Filtered", "split": "train[:250]", "trust_remote_code": false, "type": "chat_template" }, { "chat_template": "tokenizer_default", "message_property_mappings": { "content": "content", "role": "role" }, "path": "ConicCat/Ao3_Soft_Refusal", "trust_remote_code": false, "type": "chat_template" }, { "chat_template": "tokenizer_default", "message_property_mappings": { "content": "content", "role": "role" }, "path": "ConicCat/VSF", "trust_remote_code": false, "type": "chat_template" } ], "ddp": false, "device": "cuda:0", "device_map": "auto", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "eaft_alpha": 1.0, "eaft_k": 20, "env_capabilities": { "torch_version": "2.8.0" }, "eval_batch_size": 2, "eval_causal_lm_metrics": [ "sacrebleu", "comet", "ter", "chrf" ], "eval_max_new_tokens": 128, "eval_sample_packing": true, "eval_table_size": 0, "experimental_skip_move_to_device": true, "flash_attention": false, "fp16": false, "generate_samples": false, "generation_do_sample": true, "generation_max_new_tokens": 50, "generation_prompt_ratio": 0.5, "generation_temperature": 0.7, "gradient_accumulation_steps": 8, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": { "use_reentrant": true }, "include_tkps": true, "learning_rate": 2.5e-05, "liger_fused_linear_cross_entropy": true, "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": 64, "lora_dropout": 0.0, "lora_model_dir": "./Olmo-Stage1/", "lora_qkv_kernel": false, "lora_r": 32, "lora_target_linear": true, "loraplus_lr_embedding": 1e-06, "loraplus_lr_ratio": 16.0, "lr_scheduler": "constant_with_warmup", "max_grad_norm": 1.0, "mean_resizing_embeddings": false, "merge_lora": true, "micro_batch_size": 2, "model_config_type": "olmo3", "num_epochs": 3.0, "num_generation_samples": 3, "optimizer": "paged_adamw_8bit", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "./Olmo-Stage1", "pad_to_sequence_len": true, "plugins": [ "axolotl.integrations.liger.LigerPlugin" ], "pretrain_multipack_attn": true, "profiler_steps_start": 0, "qlora_sharded_model_loading": false, "quantize_moe_experts": 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_strategy": "no", "seed": 42, "sequence_len": 6144, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": true, "skip_prepare_dataset": false, "special_tokens": { "eos_token": "<|im_end|>" }, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tf32": true, "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_tensorboard": true, "val_set_size": 0.0, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "warmup_ratio": 0.05, "weight_decay": 0.0, "world_size": 1 } [2026-03-20 00:33:26,557] [INFO] [axolotl.cli.utils.load.load_model_and_tokenizer:40] [PID:13067] loading tokenizer... allenai/Olmo-3.1-32B-Instruct [2026-03-20 00:33:27,178] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:299] [PID:13067] EOS: 100265 / <|im_end|> [2026-03-20 00:33:27,178] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:300] [PID:13067] BOS: 100257 / <|endoftext|> [2026-03-20 00:33:27,178] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:301] [PID:13067] PAD: 100277 / <|pad|> [2026-03-20 00:33:27,178] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:302] [PID:13067] UNK: 100257 / <|endoftext|> [2026-03-20 00:33:27,178] [INFO] [axolotl.cli.utils.load.load_model_and_tokenizer:43] [PID:13067] loading model... [2026-03-20 00:33:27,237] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_evaluation_loop:91] [PID:13067] Patched Trainer.evaluation_loop with nanmean loss calculation [2026-03-20 00:33:27,238] [DEBUG] [axolotl.monkeypatch.transformers.trainer_loss_calc.patch_maybe_log_save_evaluate:142] [PID:13067] Patched Trainer._maybe_log_save_evaluate with nanmean loss calculation [2026-03-20 00:33:27,239] [INFO] [axolotl.loaders.patch_manager._apply_multipack_patches:389] [PID:13067] Applying multipack dataloader patch for sample packing... [2026-03-20 00:33:28,169] [INFO] [axolotl.integrations.liger.plugin.pre_model_load:104] [PID:13067] Applying LIGER to olmo3 with kwargs: {'rope': True, 'cross_entropy': None, 'fused_linear_cross_entropy': True, 'rms_norm': True, 'swiglu': True} Loading weights: 0%| | 0/707 [00:00