| { | |
| "model_config": { | |
| "name": "QwenVL", | |
| "model_path": "Qwen/Qwen2.5-VL-72B-Instruct", | |
| "trust_remote_code": true, | |
| "torch_dtype": "auto", | |
| "device_map": "auto", | |
| "low_cpu_mem_usage": true, | |
| "use_cache": false, | |
| "cache_dir": null | |
| }, | |
| "compression_config": { | |
| "name": "PTQ", | |
| "quantization": { | |
| "name": "fp8_dynamic", | |
| "bits": 8, | |
| "quant_method": { | |
| "weight": "per-tensor", | |
| "activation": "per-tensor" | |
| }, | |
| "quant_helpers": [], | |
| "smooth_alpha": 0.5, | |
| "low_memory": false, | |
| "modules_to_quantize": [], | |
| "zero_point": true, | |
| "mse_range": false, | |
| "ignore_layers": [ | |
| "model.visual.patch_embed.proj", | |
| "model.lm_head", | |
| "model.language_model.embed_tokens", | |
| "model.visual.merger.mlp.0", | |
| "model.visual.merger.mlp.2", | |
| "lm_head" | |
| ], | |
| "quant_analyse": false, | |
| "quant_vit": true | |
| }, | |
| "cache": null | |
| }, | |
| "dataset_config": null, | |
| "global_config": { | |
| "save_path": "./qwen2_5_vl-72b_fp8_dynamic", | |
| "max_seq_length": 2048, | |
| "hidden_size": 8192, | |
| "model_arch_type": "qwen2_5_vl", | |
| "deploy_backend": "vllm" | |
| }, | |
| "infer_config": null, | |
| "debug_info": { | |
| "python": "3.12.11 (main, Jun 4 2025, 08:56:18) [GCC 11.4.0]", | |
| "angelslim": { | |
| "name": "angelslim", | |
| "version": "677193e9ad36197c3a8ff1caa82dd230abf2447e", | |
| "source": "git" | |
| }, | |
| "torch": { | |
| "name": "torch", | |
| "version": "2.7.1", | |
| "source": "pip" | |
| }, | |
| "transformers": { | |
| "name": "transformers", | |
| "version": "4.53.3", | |
| "source": "pip" | |
| }, | |
| "torch_cuda_version": "12.6" | |
| } | |
| } |