[2025-11-21 19:32:32,550] [DEBUG] [axolotl.utils.config.resolve_dtype:66] [PID:1196] bf16 support detected, enabling for this configuration. config.json: 0.00B [00:00, ?B/s] config.json: 2.88kB [00:00, 14.6MB/s] [2025-11-21 19:32:32,837] [DEBUG] [axolotl.utils.config.log_gpu_memory_usage:127] [PID:1196] baseline 0.000GB () [2025-11-21 19:32:32,838] [INFO] [axolotl.cli.config.load_cfg:256] [PID:1196] config: { "accelerator_config": { "dispatch_batches": false, "split_batches": false }, "activation_offloading": false, "adam_beta1": 0.9, "adam_beta2": 0.98, "adam_epsilon": 1e-06, "axolotl_config_path": "embeddings-12b-ukr-transtokenizers.yaml", "base_model": "Goader/gemma-3-12b-pt-transtokenizers", "base_model_config": "Goader/gemma-3-12b-pt-transtokenizers", "batch_size": 8, "bf16": true, "capabilities": { "bf16": true, "compute_capability": "sm_90", "fp8": false, "n_gpu": 1, "n_node": 1 }, "context_parallel_size": 1, "dataloader_num_workers": 8, "dataloader_prefetch_factor": 9, "dataset_num_proc": 64, "dataset_prepared_path": "last_run_prepared_embeddings", "ddp": false, "ddp_find_unused_parameters": false, "ddp_timeout": 7200, "device": "cuda:0", "dion_rank_fraction": 1.0, "dion_rank_multiple_of": 1, "env_capabilities": { "torch_version": "2.8.0" }, "eval_batch_size": 4, "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": true, "fp16": false, "gradient_accumulation_steps": 2, "gradient_checkpointing": false, "include_tkps": true, "is_multimodal": true, "learning_rate": 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": 10, "lora_dropout": 0.0, "loraplus_lr_embedding": 1e-06, "lr_scheduler": "warmup_stable_decay", "lr_scheduler_kwargs": { "min_lr_ratio": 0.05, "num_decay_steps": 10000 }, "max_grad_norm": 1.0, "max_steps": 15000, "mean_resizing_embeddings": false, "micro_batch_size": 4, "model_config_type": "gemma3", "num_epochs": 1.0, "optimizer": "adamw_torch_fused", "otel_metrics_host": "localhost", "otel_metrics_port": 8000, "output_dir": "./outputs/gemma-3-12b-transtokenizers-pt", "pad_to_sequence_len": true, "plugins": [ "axolotl.integrations.liger.LigerPlugin" ], "pretrain_multipack_attn": true, "pretraining_dataset": [ { "message_property_mappings": { "content": "content", "role": "role" }, "path": "Goader/kobza-2m-jsonl", "trust_remote_code": false, "type": "pretrain" } ], "processor_config": "Goader/gemma-3-12b-pt-transtokenizers", "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_steps": 5000, "save_total_limit": 30, "sequence_len": 1024, "shuffle_before_merging_datasets": false, "shuffle_merged_datasets": false, "skip_prepare_dataset": false, "streaming_multipack_buffer_size": 10000, "strict": false, "tensor_parallel_size": 1, "tiled_mlp_use_original_mlp": true, "tokenizer_config": "lapa-llm/tokenizer", "tokenizer_save_jinja_files": true, "torch_dtype": "torch.bfloat16", "train_on_inputs": true, "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 }, "unfrozen_parameters": [ "^lm_head.weight$", "^model.language_model.embed_tokens.weight$" ], "use_otel_metrics": false, "use_ray": false, "use_wandb": true, "val_set_size": 0.0, "vllm": { "device": "auto", "dtype": "auto", "gpu_memory_utilization": 0.9, "host": "0.0.0.0", "port": 8000 }, "wandb_project": "matt", "warmup_ratio": 0.1, "weight_decay": 0.01, "world_size": 1 } [2025-11-21 19:32:33,114] [DEBUG] [axolotl.loaders.utils.check_model_config:83] [PID:1196] Loaded image size: 896 from model config tokenizer_config.json: 0.00B [00:00, ?B/s] tokenizer_config.json: 52.7kB [00:00, 30.6MB/s] tokenizer.json: 0%| | 0.00/37.4M [00:00 [2025-11-21 19:32:37,916] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1196] BOS: 2 / [2025-11-21 19:32:37,916] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:282] [PID:1196] PAD: 0 / [2025-11-21 19:32:37,916] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:283] [PID:1196] UNK: 3 / [2025-11-21 19:32:40,076] [DEBUG] [axolotl.utils.data.streaming.wrap_streaming_dataset:231] [PID:1196] NOT shuffling merged pretraining datasets [2025-11-21 19:32:41,367] [DEBUG] [axolotl.train.setup_model_and_tokenizer:70] [PID:1196] loading tokenizer... lapa-llm/tokenizer [2025-11-21 19:32:42,716] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:280] [PID:1196] EOS: 1 / [2025-11-21 19:32:42,716] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:281] [PID:1196] BOS: 2 / [2025-11-21 19:32:42,716] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:282] [PID:1196] PAD: 0 / [2025-11-21 19:32:42,716] [DEBUG] [axolotl.loaders.tokenizer.load_tokenizer:283] [PID:1196] UNK: 3 / processor_config.json: 0%| | 0.00/70.0 [00:00