ddz16/CamInject-4B

1{2  "per_device_train_batch_size": 2,3  "num_train_epochs": 2.0,4  "max_steps": -1,5  "learning_rate": 2e-05,6  "lr_scheduler_type": "cosine",7  "lr_scheduler_kwargs": null,8  "warmup_steps": 0,9  "optim": "adamw_torch_fused",10  "optim_args": null,11  "weight_decay": 0.01,12  "adam_beta1": 0.9,13  "adam_beta2": 0.95,14  "adam_epsilon": 1e-08,15  "optim_target_modules": null,16  "gradient_accumulation_steps": 4,17  "average_tokens_across_devices": true,18  "max_grad_norm": 1.0,19  "label_smoothing_factor": 0.0,20  "bf16": true,21  "fp16": false,22  "bf16_full_eval": false,23  "fp16_full_eval": false,24  "tf32": null,25  "gradient_checkpointing": true,26  "gradient_checkpointing_kwargs": null,27  "torch_compile": false,28  "torch_compile_backend": null,29  "torch_compile_mode": null,30  "use_liger_kernel": false,31  "liger_kernel_config": null,32  "use_cache": false,33  "neftune_noise_alpha": null,34  "torch_empty_cache_steps": null,35  "auto_find_batch_size": false,36  "logging_strategy": "steps",37  "logging_steps": 10,38  "logging_first_step": true,39  "log_on_each_node": true,40  "logging_nan_inf_filter": true,41  "include_num_input_tokens_seen": false,42  "log_level": "passive",43  "log_level_replica": "warning",44  "disable_tqdm": null,45  "report_to": [46    "wandb"47  ],48  "project": "huggingface",49  "trackio_space_id": null,50  "trackio_bucket_id": null,51  "trackio_static_space_id": null,52  "eval_strategy": "no",53  "eval_steps": null,54  "eval_delay": 0,55  "per_device_eval_batch_size": 2,56  "prediction_loss_only": false,57  "eval_on_start": false,58  "eval_do_concat_batches": true,59  "eval_use_gather_object": false,60  "eval_accumulation_steps": null,61  "include_for_metrics": [],62  "batch_eval_metrics": false,63  "save_only_model": false,64  "save_strategy": "epoch",65  "save_steps": 500,66  "save_on_each_node": false,67  "save_total_limit": 1,68  "enable_jit_checkpoint": false,69  "push_to_hub": false,70  "hub_token": null,71  "hub_private_repo": null,72  "hub_model_id": null,73  "hub_strategy": "every_save",74  "hub_always_push": false,75  "hub_revision": null,76  "load_best_model_at_end": false,77  "metric_for_best_model": "loss",78  "greater_is_better": false,79  "ignore_data_skip": false,80  "restore_callback_states_from_checkpoint": false,81  "full_determinism": false,82  "seed": 42,83  "data_seed": 42,84  "use_cpu": false,85  "accelerator_config": {86    "dispatch_batches": false87  },88  "parallelism_config": null,89  "dataloader_drop_last": false,90  "dataloader_num_workers": 8,91  "dataloader_pin_memory": true,92  "dataloader_persistent_workers": false,93  "dataloader_prefetch_factor": null,94  "remove_unused_columns": true,95  "label_names": null,96  "train_sampling_strategy": "random",97  "length_column_name": "length",98  "ddp_find_unused_parameters": null,99  "ddp_bucket_cap_mb": null,100  "ddp_broadcast_buffers": null,101  "ddp_static_graph": null,102  "ddp_backend": null,103  "ddp_timeout": 18000000,104  "fsdp": [],105  "fsdp_config": null,106  "deepspeed": {107    "fp16": {108      "enabled": "auto",109      "loss_scale": 0,110      "loss_scale_window": 1000,111      "initial_scale_power": 16,112      "hysteresis": 2,113      "min_loss_scale": 1114    },115    "bf16": {116      "enabled": "auto"117    },118    "zero_optimization": {119      "stage": 2,120      "offload_optimizer": {121        "device": "none",122        "pin_memory": true123      },124      "allgather_partitions": true,125      "allgather_bucket_size": 200000000.0,126      "overlap_comm": false,127      "reduce_scatter": true,128      "reduce_bucket_size": 200000000.0,129      "contiguous_gradients": true130    },131    "gradient_accumulation_steps": "auto",132    "gradient_clipping": "auto",133    "steps_per_print": 2000,134    "train_batch_size": "auto",135    "train_micro_batch_size_per_gpu": "auto",136    "wall_clock_breakdown": false137  },138  "debug": null,139  "skip_memory_metrics": true,140  "do_train": false,141  "do_eval": false,142  "do_predict": false,143  "resume_from_checkpoint": null,144  "warmup_ratio": 0.05,145  "local_rank": 0,146  "sortish_sampler": false,147  "predict_with_generate": false,148  "generation_max_length": null,149  "generation_num_beams": null,150  "generation_config": null,151  "tuner_backend": "peft",152  "vit_gradient_checkpointing": false,153  "router_aux_loss_coef": 0.0,154  "enable_dft_loss": false,155  "enable_channel_loss": false,156  "safe_serialization": true,157  "max_shard_size": "5GB",158  "check_model": true,159  "acc_strategy": "token",160  "train_dataloader_shuffle": true,161  "group_by_length": false,162  "max_epochs": null,163  "aligner_lr": null,164  "vit_lr": null,165  "use_logits_to_keep": null,166  "ds3_gather_for_generation": true,167  "resume_only_model": false,168  "optimizer": null,169  "loss_type": null,170  "eval_metric": null,171  "callbacks": [],172  "early_stop_interval": null,173  "eval_use_evalscope": false,174  "eval_dataset": [],175  "eval_dataset_args": null,176  "eval_limit": null,177  "eval_generation_config": null,178  "extra_eval_args": null,179  "tuner_type": "full",180  "use_galore": false,181  "galore_target_modules": null,182  "galore_rank": 128,183  "galore_update_proj_gap": 50,184  "galore_scale": 1.0,185  "galore_proj_type": "std",186  "galore_optim_per_parameter": false,187  "galore_with_embedding": false,188  "galore_quantization": false,189  "galore_proj_quant": false,190  "galore_proj_bits": 4,191  "galore_proj_group_size": 256,192  "galore_cos_threshold": 0.4,193  "galore_gamma_proj": 2,194  "galore_queue_size": 5,195  "lisa_activated_layers": 0,196  "lisa_step_interval": 20,197  "use_flash_ckpt": false,198  "use_ray": false,199  "ray_exp_name": null,200  "device_groups": null,201  "model": "Qwen/Qwen3-VL-4B-Instruct",202  "model_type": "qwen3_vl_caminject",203  "model_revision": null,204  "task_type": "causal_lm",205  "torch_dtype": "bfloat16",206  "attn_impl": "flash_attention_2",207  "experts_impl": null,208  "new_special_tokens": [],209  "num_labels": null,210  "problem_type": null,211  "rope_scaling": null,212  "device_map": null,213  "max_memory": {},214  "max_model_len": null,215  "local_repo_path": null,216  "init_strategy": null,217  "template": "qwen3_vl",218  "system": null,219  "max_length": 16384,220  "truncation_strategy": "delete",221  "max_pixels": null,222  "agent_template": null,223  "norm_bbox": null,224  "use_chat_template": true,225  "padding_side": "right",226  "padding_free": false,227  "loss_scale": "default",228  "sequence_parallel_size": 1,229  "template_backend": "swift",230  "response_prefix": null,231  "enable_thinking": null,232  "preserve_thinking": null,233  "add_non_thinking_prefix": true,234  "val_dataset": [],235  "cached_dataset": [],236  "cached_val_dataset": [],237  "split_dataset_ratio": 0.0,238  "dataset_num_proc": 8,239  "load_from_cache_file": true,240  "dataset_shuffle": true,241  "val_dataset_shuffle": false,242  "streaming": false,243  "interleave_prob": null,244  "stopping_strategy": "first_exhausted",245  "shuffle_buffer_size": 1000,246  "download_mode": "reuse_dataset_if_exists",247  "columns": {},248  "strict": false,249  "disable_auto_column_mapping": false,250  "model_name": null,251  "model_author": null,252  "custom_dataset_info": [],253  "quant_method": null,254  "quant_bits": null,255  "hqq_axis": null,256  "bnb_4bit_compute_dtype": "bfloat16",257  "bnb_4bit_quant_type": "nf4",258  "bnb_4bit_use_double_quant": true,259  "bnb_4bit_quant_storage": null,260  "max_new_tokens": 64,261  "temperature": 0.0,262  "top_k": null,263  "top_p": null,264  "repetition_penalty": null,265  "num_beams": 1,266  "stream": false,267  "stop_words": [],268  "logprobs": false,269  "top_logprobs": null,270  "structured_outputs_regex": null,271  "adapters": [],272  "external_plugins": [273    "camera_movement_sft/plugins/camdistill_plugin.py"274  ],275  "custom_register_path": [],276  "model_kwargs": {},277  "enable_npu_model_patch": true,278  "load_args": false,279  "load_data_args": false,280  "packing": false,281  "packing_length": null,282  "packing_num_proc": 1,283  "lazy_tokenize": true,284  "use_hf": true,285  "ignore_args_error": false,286  "use_swift_lora": false,287  "freeze_parameters": [],288  "freeze_parameters_regex": null,289  "freeze_parameters_ratio": 0.0,290  "trainable_parameters": [],291  "trainable_parameters_regex": null,292  "freeze_llm": false,293  "freeze_vit": true,294  "freeze_aligner": true,295  "target_modules": [296    "all-linear"297  ],298  "target_regex": null,299  "target_parameters": null,300  "modules_to_save": [],301  "lora_rank": 8,302  "lora_alpha": 32,303  "lora_dropout": 0.05,304  "lora_bias": "none",305  "lora_dtype": null,306  "lorap_lr_ratio": null,307  "use_rslora": false,308  "use_dora": false,309  "lora_ga_batch_size": 2,310  "lora_ga_iters": 2,311  "lora_ga_max_length": 1024,312  "lora_ga_direction": "ArB2r",313  "lora_ga_scale": "stable",314  "lora_ga_stable_gamma": 16,315  "init_weights": true,316  "fourier_n_frequency": 2000,317  "fourier_scaling": 300.0,318  "boft_block_size": 4,319  "boft_block_num": 0,320  "boft_n_butterfly_factor": 1,321  "boft_dropout": 0.0,322  "vera_rank": 256,323  "vera_projection_prng_key": 0,324  "vera_dropout": 0.0,325  "vera_d_initial": 0.1,326  "adapter_act": "gelu",327  "adapter_length": 128,328  "adalora_target_r": 8,329  "adalora_init_r": 12,330  "adalora_tinit": 0,331  "adalora_tfinal": 0,332  "adalora_deltaT": 1,333  "adalora_beta1": 0.85,334  "adalora_beta2": 0.85,335  "adalora_orth_reg_weight": 0.5,336  "llamapro_num_new_blocks": 4,337  "llamapro_num_groups": null,338  "reft_layer_key": null,339  "reft_layers": null,340  "reft_rank": 4,341  "reft_intervention_type": "LoreftIntervention",342  "reft_args": null,343  "swanlab_token": null,344  "swanlab_project": "ms-swift",345  "swanlab_workspace": null,346  "swanlab_exp_name": null,347  "swanlab_notification_method": null,348  "swanlab_webhook_url": null,349  "swanlab_secret": null,350  "swanlab_sender_email": null,351  "swanlab_receiver_email": null,352  "swanlab_smtp_server": null,353  "swanlab_smtp_port": null,354  "swanlab_email_language": "zh",355  "swanlab_mode": "cloud",356  "add_version": true,357  "create_checkpoint_symlink": false,358  "zero_hpz_partition_size": null,359  "deepspeed_autotp_size": null,360  "swift_version": "4.3.0.dev0",361  "ckpt_dir": null,362  "rank": 0,363  "global_world_size": 8,364  "local_world_size": 8,365  "model_suffix": "Qwen3-VL-4B-Instruct",366  "template_meta": "QwenTemplateMeta(template_type='qwen3_vl', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3VLTemplate'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='hermes', is_thinking=False, thinking_prefix='<think>\\n', non_thinking_prefix='', history_thinking_prefix='')",367  "_val_dataset_exists": true,368  "hub": "<class 'swift.hub.hub.HFHub'>",369  "evaluation_strategy": "no"370}