{ "output_dir": "/root/autodl-tmp/output_grpo/v2-20250628-104314", "overwrite_output_dir": false, "do_train": false, "do_eval": false, "do_predict": false, "eval_strategy": "steps", "prediction_loss_only": false, "per_device_train_batch_size": 1, "per_device_eval_batch_size": 4, "per_gpu_train_batch_size": null, "per_gpu_eval_batch_size": null, "gradient_accumulation_steps": 4, "eval_accumulation_steps": null, "eval_delay": 0, "torch_empty_cache_steps": null, "learning_rate": 1e-06, "weight_decay": 0.1, "adam_beta1": 0.9, "adam_beta2": 0.95, "adam_epsilon": 1e-08, "max_grad_norm": 1.0, "num_train_epochs": 1.0, "max_steps": 15000, "lr_scheduler_type": "cosine", "lr_scheduler_kwargs": null, "warmup_ratio": 0.01, "warmup_steps": 0, "log_level": "passive", "log_level_replica": "warning", "log_on_each_node": true, "logging_dir": "/root/autodl-tmp/output_grpo/v2-20250628-104314/runs", "logging_strategy": "steps", "logging_first_step": true, "logging_steps": 1, "logging_nan_inf_filter": true, "save_strategy": "steps", "save_steps": 500.0, "save_total_limit": 1, "save_safetensors": true, "save_on_each_node": false, "save_only_model": false, "restore_callback_states_from_checkpoint": false, "no_cuda": false, "use_cpu": false, "use_mps_device": false, "seed": 42, "data_seed": 42, "jit_mode_eval": false, "use_ipex": false, "bf16": true, "fp16": false, "fp16_opt_level": "O1", "half_precision_backend": "auto", "bf16_full_eval": false, "fp16_full_eval": false, "tf32": null, "local_rank": 0, "ddp_backend": null, "tpu_num_cores": null, "tpu_metrics_debug": false, "debug": null, "dataloader_drop_last": false, "eval_steps": 500.0, "dataloader_num_workers": 4, "dataloader_prefetch_factor": null, "past_index": -1, "run_name": "/root/autodl-tmp/output_grpo/v2-20250628-104314", "disable_tqdm": null, "remove_unused_columns": false, "label_names": null, "load_best_model_at_end": false, "metric_for_best_model": "reward", "greater_is_better": true, "ignore_data_skip": false, "fsdp": "", "fsdp_min_num_params": 0, "fsdp_config": null, "tp_size": 0, "fsdp_transformer_layer_cls_to_wrap": null, "accelerator_config": { "dispatch_batches": false }, "deepspeed": { "fp16": { "enabled": "auto", "loss_scale": 0, "loss_scale_window": 1000, "initial_scale_power": 16, "hysteresis": 2, "min_loss_scale": 1 }, "bf16": { "enabled": "auto" }, "zero_optimization": { "stage": 3, "offload_optimizer": { "device": "none", "pin_memory": true }, "offload_param": { "device": "none", "pin_memory": true }, "overlap_comm": false, "contiguous_gradients": true, "sub_group_size": 1000000000.0, "reduce_bucket_size": "auto", "zero_quantized_weights": false, "zero_quantized_gradients": false, "stage3_prefetch_bucket_size": "auto", "stage3_param_persistence_threshold": "auto", "stage3_max_live_parameters": 1000000000.0, "stage3_max_reuse_distance": 1000000000.0, "stage3_gather_16bit_weights_on_model_save": true }, "gradient_accumulation_steps": "auto", "gradient_clipping": "auto", "steps_per_print": 2000, "train_batch_size": "auto", "train_micro_batch_size_per_gpu": "auto", "wall_clock_breakdown": false }, "label_smoothing_factor": 0.0, "optim": "adamw_torch", "optim_args": null, "adafactor": false, "group_by_length": false, "length_column_name": "length", "report_to": [ "swanlab" ], "ddp_find_unused_parameters": null, "ddp_bucket_cap_mb": null, "ddp_broadcast_buffers": null, "dataloader_pin_memory": true, "dataloader_persistent_workers": false, "skip_memory_metrics": true, "use_legacy_prediction_loop": false, "push_to_hub": false, "resume_from_checkpoint": null, "hub_model_id": null, "hub_strategy": "every_save", "hub_token": null, "hub_private_repo": null, "hub_always_push": false, "gradient_checkpointing": true, "gradient_checkpointing_kwargs": null, "include_inputs_for_metrics": false, "include_for_metrics": [], "eval_do_concat_batches": true, "fp16_backend": "auto", "push_to_hub_model_id": null, "push_to_hub_organization": null, "push_to_hub_token": null, "mp_parameters": "", "auto_find_batch_size": false, "full_determinism": false, "torchdynamo": null, "ray_scope": "last", "ddp_timeout": 18000000, "torch_compile": false, "torch_compile_backend": null, "torch_compile_mode": null, "include_tokens_per_second": false, "include_num_input_tokens_seen": false, "neftune_noise_alpha": null, "optim_target_modules": null, "batch_eval_metrics": false, "eval_on_start": false, "use_liger_kernel": false, "eval_use_gather_object": false, "average_tokens_across_devices": false, "sortish_sampler": false, "predict_with_generate": false, "generation_max_length": null, "generation_num_beams": null, "generation_config": null, "vit_gradient_checkpointing": null, "check_model": true, "acc_strategy": "token", "train_dataloader_shuffle": true, "max_epochs": null, "aligner_lr": null, "vit_lr": null, "optimizer": null, "use_logits_to_keep": null, "channels": null, "ds3_gather_for_generation": true, "metric_warmup_step": 0, "fsdp_num": 1, "acc_steps": 1, "eval_use_evalscope": false, "eval_datasets": [], "eval_limit": null, "eval_datasets_args": null, "eval_generation_config": null, "model": "/root/autodl-tmp/landai_0610_sft", "model_type": "qwen2_5_vl", "model_revision": null, "task_type": "causal_lm", "torch_dtype": "bfloat16", "attn_impl": "flash_attn", "num_labels": null, "problem_type": null, "rope_scaling": null, "device_map": null, "max_memory": {}, "local_repo_path": null, "init_strategy": null, "template": "qwen2_5_vl", "system": "You are a top-level AI remote sensing analysis expert. Core principles: visual priority, data fusion. Your analysis must start from the visual evidence of remote sensing images. Use the provided Points of Interest (POI) and temporal density data as key inputs to refine and determine the final category. You must explicitly explain how you weigh these pieces of evidence. Response workflow:\\n1. Stage: Data extraction and structured analysis; Plan: Briefly describe your plan, including how to parse and classify the raw POI strings and analyze temporal data patterns. Structured output: Output a JSON containing your analysis of each data source. The JSON for POI and temporal data should include your analysis of the raw data, not just the raw data itself. For example: {\"visual_evidence\": [{\"bbox_2d\": [x,y,x,y], \"label\": \"e.g., dense building clusters, road networks, green spaces\"}], \"poi_analysis\": {\"dominant_category_summary\": [{\"primary_category\": \"Residential\", \"indicative_poi_count\": 57, \"examples\": [\"Residential Community-57\"]}, {\"primary_category\": \"Commercial\", \"indicative_poi_count\": 486, \"examples\": [\"Other Dining-101\", \"Shoe Store-108\", \"Other Retail-70\", \"Convenience Store-28\", \"Chinese Restaurant-37\"]}, {\"primary_category\": \"Public Service\", \"indicative_poi_count\": 21, \"examples\": [\"Training Center-13\", \"Kindergarten-8\"]}], \"interpretation\": \"Clear mix of commercial and residential, with a large residential base.\"}, \"temporal_analysis\": {\"pattern_identified\": \"Bimodal pattern: smaller peak in the morning, larger and sustained peak in the evening/night.\", \"peak_hours\": \"18:00-22:00\", \"interpretation\": \"This pattern is typical of dense residential areas where residents return home after work and engage in local commercial activities.\"}}. You must enclose all content of this stage within .\\n2. Stage: Synthesis and argumentation; Synthesize all findings. Start from visual evidence and its limitations. The key is to weigh POI and temporal evidence to resolve ambiguities. For example, argue why a higher number of commercial POIs might still indicate a residential area, or vice versa. Finally, determine the most predominant functional category through reasoning. You must enclose all content of this stage within .\\n3. Stage: Final classification; Provide a single most appropriate category from the list. Enclose the final answer within tags. \n", "max_length": 12000, "truncation_strategy": "left", "max_pixels": null, "agent_template": null, "norm_bbox": null, "use_chat_template": true, "padding_free": true, "padding_side": "right", "loss_scale": "last_round", "sequence_parallel_size": 1, "response_prefix": null, "template_backend": "swift", "dataset": [ "/root/autodl-tmp/mixed_dapo_train_24376s_add_solution.jsonl" ], "val_dataset": [], "split_dataset_ratio": 0.01, "dataset_num_proc": 1, "load_from_cache_file": true, "dataset_shuffle": true, "val_dataset_shuffle": false, "streaming": false, "interleave_prob": null, "stopping_strategy": "first_exhausted", "shuffle_buffer_size": 1000, "download_mode": "reuse_dataset_if_exists", "columns": {}, "strict": false, "model_name": null, "model_author": null, "custom_dataset_info": [], "quant_method": null, "quant_bits": null, "hqq_axis": null, "bnb_4bit_compute_dtype": "bfloat16", "bnb_4bit_quant_type": "nf4", "bnb_4bit_use_double_quant": true, "bnb_4bit_quant_storage": null, "max_new_tokens": 8000, "temperature": 1.0, "top_k": null, "top_p": null, "repetition_penalty": 1.1, "num_beams": 1, "stream": false, "stop_words": [], "logprobs": false, "top_logprobs": null, "ckpt_dir": "/root/autodl-tmp/landai_0610_sft", "lora_modules": [], "tuner_backend": "peft", "train_type": "full", "adapters": [], "external_plugins": [ ], "model_kwargs": {}, "load_args": false, "load_data_args": false, "packing": false, "packing_cache": null, "custom_register_path": [], "use_hf": false, "ignore_args_error": false, "use_swift_lora": false, "freeze_parameters": [ "visual", "visual.merger" ], "freeze_parameters_regex": null, "freeze_parameters_ratio": 0.0, "trainable_parameters": [], "trainable_parameters_regex": null, "freeze_llm": false, "freeze_vit": true, "freeze_aligner": true, "target_modules": [ "all-linear" ], "target_regex": null, "modules_to_save": [], "lora_rank": 8, "lora_alpha": 32, "lora_dropout": 0.05, "lora_bias": "none", "lora_dtype": null, "lorap_lr_ratio": null, "use_rslora": false, "use_dora": false, "lora_ga_batch_size": 2, "lora_ga_iters": 2, "lora_ga_max_length": 1024, "lora_ga_direction": "ArB2r", "lora_ga_scale": "stable", "lora_ga_stable_gamma": 16, "init_weights": true, "fourier_n_frequency": 2000, "fourier_scaling": 300.0, "boft_block_size": 4, "boft_block_num": 0, "boft_n_butterfly_factor": 1, "boft_dropout": 0.0, "vera_rank": 256, "vera_projection_prng_key": 0, "vera_dropout": 0.0, "vera_d_initial": 0.1, "adapter_act": "gelu", "adapter_length": 128, "use_galore": false, "galore_target_modules": null, "galore_rank": 128, "galore_update_proj_gap": 50, "galore_scale": 1.0, "galore_proj_type": "std", "galore_optim_per_parameter": false, "galore_with_embedding": false, "galore_quantization": false, "galore_proj_quant": false, "galore_proj_bits": 4, "galore_proj_group_size": 256, "galore_cos_threshold": 0.4, "galore_gamma_proj": 2, "galore_queue_size": 5, "adalora_target_r": 8, "adalora_init_r": 12, "adalora_tinit": 0, "adalora_tfinal": 0, "adalora_deltaT": 1, "adalora_beta1": 0.85, "adalora_beta2": 0.85, "adalora_orth_reg_weight": 0.5, "llamapro_num_new_blocks": 4, "llamapro_num_groups": null, "lisa_activated_layers": 0, "lisa_step_interval": 20, "reft_layer_key": null, "reft_layers": null, "reft_rank": 4, "reft_intervention_type": "LoreftIntervention", "reft_args": null, "swanlab_token": null, "swanlab_project": null, "swanlab_workspace": null, "swanlab_exp_name": "/root/autodl-tmp/output_grpo/v2-20250628-104314", "swanlab_mode": "cloud", "add_version": true, "resume_only_model": false, "create_checkpoint_symlink": false, "lazy_tokenize": true, "loss_type": "grpo", "metric": null, "zero_hpz_partition_size": null, "sft_alpha": 0, "reward_model": null, "reward_adapters": [], "reward_model_type": null, "reward_model_revision": null, "num_ppo_epochs": 4, "whiten_rewards": false, "kl_coef": 0.05, "cliprange": 0.2, "vf_coef": 0.1, "cliprange_value": 0.2, "gamma": 1.0, "lam": 0.95, "num_mini_batches": 1, "local_rollout_forward_batch_size": 64, "num_sample_generations": 10, "response_length": 8000, "missing_eos_penalty": null, "epsilon": 0.2, "epsilon_high": null, "delta": null, "num_infer_workers": null, "vllm_mode": "server", "vllm_device": null, "vllm_gpu_memory_utilization": 0.9, "vllm_max_model_len": null, "vllm_max_num_seqs": null, "vllm_enforce_eager": false, "vllm_limit_mm_per_prompt": null, "vllm_enable_prefix_caching": true, "vllm_tensor_parallel_size": 1, "vllm_server_base_url": null, "vllm_server_host": "127.0.0.1", "vllm_server_port": 8000, "vllm_server_timeout": 240.0, "cosine_min_len_value_wrong": -0.5, "cosine_max_len_value_wrong": 0.0, "cosine_min_len_value_correct": 1.0, "cosine_max_len_value_correct": 0.5, "cosine_max_len": null, "repetition_n_grams": 3, "repetition_max_penalty": -1.0, "reward_model_plugin": null, "sync_ref_model": false, "ref_model_sync_steps": 512, "ref_model_mixup_alpha": 0.6, "async_generate": false, "tensor_parallel_size": null, "sleep_level": 0, "move_model_batches": null, "offload_optimizer": false, "offload_model": false, "gc_collect_after_offload": false, "multi_turn_func": null, "multi_turn_scheduler": null, "max_turns": null, "completion_length_limit_scope": "per_round", "dynamic_sample": false, "max_resample_times": 3, "overlong_filter": true, "soft_max_length": null, "soft_cache_length": null, "scale_rewards": true, "wandb_log_unique_prompts": null, "generation_batch_size": null, "steps_per_generation": null, "num_generations": 12, "reward_funcs": [ "custom_format_correctness", "custom_answer_accuracy" ], "reward_weights": [ 1.0, 1.0 ], "log_completions": true, "use_vllm": true, "num_iterations": 1, "teacher_model": null, "teacher_adapters": [], "teacher_model_type": null, "teacher_model_revision": null, "rlhf_type": "grpo", "ref_model": "/root/autodl-tmp/landai_0610_sft", "ref_model_type": "qwen2_5_vl", "ref_model_revision": null, "beta": 0.04, "label_smoothing": 0, "max_completion_length": 8000, "rpo_alpha": 1.0, "cpo_alpha": 1.0, "simpo_gamma": 1, "desirable_weight": 1.0, "undesirable_weight": 1.0, "center_rewards_coefficient": null, "lmbda": 0.5, "seq_kd": false, "vllm_client": "", "rank": 0, "global_world_size": 3, "local_world_size": 3, "model_suffix": "landai_0610_sft", "model_info": "ModelInfo(model_type='qwen2_5_vl', model_dir='/root/autodl-tmp/landai_0610_sft', torch_dtype=torch.bfloat16, max_model_len=128000, quant_method=None, quant_bits=None, rope_scaling={'mrope_section': [16, 24, 24], 'rope_type': 'default', 'type': 'default'}, config=None, task_type='causal_lm', num_labels=None)", "model_meta": "ModelMeta(model_type='qwen2_5_vl', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-3B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-7B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-32B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', hf_model_id='Qwen/Qwen2.5-VL-72B-Instruct-AWQ', model_path=None, ms_revision=None, hf_revision=None)], ignore_patterns=None, requires=None, tags=[])], template='qwen2_5_vl', get_function=, model_arch='qwen2_vl', architectures=['Qwen2_5_VLForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=4.49', 'qwen_vl_utils>=0.0.6', 'decord'], tags=['vision', 'video'])", "model_dir": "/root/autodl-tmp/landai_0610_sft", "hub": "", "evaluation_strategy": "steps", "training_args": "GRPOConfig(output_dir='/root/autodl-tmp/output_grpo/v2-20250628-104314', overwrite_output_dir=False, do_train=False, do_eval=True, do_predict=False, eval_strategy=, prediction_loss_only=False, per_device_train_batch_size=1, per_device_eval_batch_size=4, per_gpu_train_batch_size=None, per_gpu_eval_batch_size=None, gradient_accumulation_steps=4, eval_accumulation_steps=None, eval_delay=0, torch_empty_cache_steps=None, learning_rate=1e-06, weight_decay=0.1, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, max_grad_norm=1.0, num_train_epochs=1.0, max_steps=15000, lr_scheduler_type=, lr_scheduler_kwargs=None, warmup_ratio=0.01, warmup_steps=0, log_level='passive', log_level_replica='warning', log_on_each_node=True, logging_dir='/root/autodl-tmp/output_grpo/v2-20250628-104314/runs', logging_strategy=, logging_first_step=True, logging_steps=1, logging_nan_inf_filter=True, save_strategy=, save_steps=500, save_total_limit=1, save_safetensors=True, save_on_each_node=False, save_only_model=False, restore_callback_states_from_checkpoint=False, no_cuda=False, use_cpu=False, use_mps_device=False, seed=42, data_seed=42, jit_mode_eval=False, use_ipex=False, bf16=True, fp16=False, fp16_opt_level='O1', half_precision_backend='auto', bf16_full_eval=False, fp16_full_eval=False, tf32=None, local_rank=0, ddp_backend=None, tpu_num_cores=None, tpu_metrics_debug=False, debug=[], dataloader_drop_last=True, eval_steps=500, dataloader_num_workers=4, dataloader_prefetch_factor=10, past_index=-1, run_name='/root/autodl-tmp/output_grpo/v2-20250628-104314', disable_tqdm=False, remove_unused_columns=False, label_names=None, load_best_model_at_end=False, metric_for_best_model='reward', greater_is_better=True, ignore_data_skip=False, fsdp=[], fsdp_min_num_params=0, fsdp_config={'min_num_params': 0, 'xla': False, 'xla_fsdp_v2': False, 'xla_fsdp_grad_ckpt': False}, tp_size=0, fsdp_transformer_layer_cls_to_wrap=None, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 0, 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, label_smoothing_factor=0.0, optim=, optim_args=None, adafactor=False, group_by_length=False, length_column_name='length', report_to=['swanlab'], ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, dataloader_pin_memory=True, dataloader_persistent_workers=False, skip_memory_metrics=True, use_legacy_prediction_loop=False, push_to_hub=False, resume_from_checkpoint=None, hub_model_id=None, hub_strategy=, hub_token=None, hub_private_repo=None, hub_always_push=False, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, include_inputs_for_metrics=False, include_for_metrics=[], eval_do_concat_batches=True, fp16_backend='auto', push_to_hub_model_id=None, push_to_hub_organization=None, push_to_hub_token=None, mp_parameters='', auto_find_batch_size=False, full_determinism=False, torchdynamo=None, ray_scope='last', ddp_timeout=18000000, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, include_tokens_per_second=None, include_num_input_tokens_seen=None, neftune_noise_alpha=None, optim_target_modules=None, batch_eval_metrics=False, eval_on_start=False, use_liger_kernel=False, eval_use_gather_object=False, average_tokens_across_devices=None, model_init_kwargs=None, disable_dropout=False, max_prompt_length=512, num_generations=12, max_completion_length=8000, ds3_gather_for_generation=True, shuffle_dataset=True, generation_batch_size=12, steps_per_generation=4, temperature=1.0, top_p=0.9, top_k=50, min_p=None, generation_kwargs=None, repetition_penalty=1.1, cache_implementation=None, use_vllm=True, vllm_server_base_url=None, vllm_mode='server', vllm_guided_decoding_regex=None, vllm_server_host='127.0.0.1', vllm_server_port=8000, vllm_server_timeout=240.0, vllm_gpu_memory_utilization=0.9, vllm_tensor_parallel_size=1, beta=0.04, num_iterations=1, epsilon=0.2, delta=None, epsilon_high=None, reward_weights=[1.0, 1.0], scale_rewards=True, loss_type='grpo', mask_truncated_completions=False, sync_ref_model=False, ref_model_mixup_alpha=0.6, ref_model_sync_steps=512, use_liger_loss=False, log_completions=True, num_completions_to_print=None, wandb_log_unique_prompts=None, vit_gradient_checkpointing=True, check_model=True, acc_strategy='token', train_dataloader_shuffle=True, max_epochs=None, aligner_lr=None, vit_lr=None, optimizer=None, use_logits_to_keep=None, channels=None, metric_warmup_step=0, fsdp_num=1, acc_steps=1, eval_use_evalscope=False, eval_datasets=[], eval_limit=None, eval_datasets_args=None, eval_generation_config=None, sft_alpha=0, train_type='full', local_repo_path=None, galore_config=None, num_infer_workers=None, vllm_device=None, vllm_max_model_len=None, vllm_max_num_seqs=None, vllm_enforce_eager=False, vllm_limit_mm_per_prompt={}, vllm_enable_prefix_caching=True, cosine_min_len_value_wrong=-0.5, cosine_max_len_value_wrong=0.0, cosine_min_len_value_correct=1.0, cosine_max_len_value_correct=0.5, cosine_max_len=8000, repetition_n_grams=3, repetition_max_penalty=-1.0, reward_model=None, reward_model_plugin=None, async_generate=False, tensor_parallel_size=None, sleep_level=0, move_model_batches=None, offload_optimizer=False, offload_model=False, gc_collect_after_offload=False, multi_turn_func=None, multi_turn_scheduler=None, max_turns=None, completion_length_limit_scope='per_round', dynamic_sample=False, max_resample_times=3, overlong_filter=True, soft_max_length=None, soft_cache_length=None, dataset_shuffle=True, stop_words=[])" }