| { | |
| "architectures": [ | |
| "MistralForCausalLM" | |
| ], | |
| "attention_dropout": 0.0, | |
| "bos_token_id": 1, | |
| "eos_token_id": 2, | |
| "head_dim": 128, | |
| "hidden_act": "silu", | |
| "hidden_size": 5120, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 14336, | |
| "max_position_embeddings": 131072, | |
| "model_type": "mistral", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 40, | |
| "num_key_value_heads": 8, | |
| "quantization_config": { | |
| "algo_config": null, | |
| "exclude": [ | |
| "lm_head" | |
| ], | |
| "export": { | |
| "kv_cache_group": [ | |
| "*k_proj", | |
| "*v_proj" | |
| ], | |
| "min_kv_scale": 0.0, | |
| "pack_method": "reorder", | |
| "weight_format": "real_quantized", | |
| "weight_merge_groups": null | |
| }, | |
| "global_quant_config": { | |
| "bias": null, | |
| "input_tensors": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| }, | |
| "output_tensors": null, | |
| "target_device": null, | |
| "weight": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| } | |
| }, | |
| "layer_quant_config": { | |
| "*k_proj": { | |
| "bias": null, | |
| "input_tensors": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| }, | |
| "output_tensors": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| }, | |
| "target_device": null, | |
| "weight": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| } | |
| }, | |
| "*v_proj": { | |
| "bias": null, | |
| "input_tensors": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| }, | |
| "output_tensors": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| }, | |
| "target_device": null, | |
| "weight": { | |
| "ch_axis": null, | |
| "dtype": "fp8_e4m3", | |
| "group_size": null, | |
| "is_dynamic": false, | |
| "observer_cls": "PerTensorMinMaxObserver", | |
| "qscheme": "per_tensor", | |
| "round_method": null, | |
| "scale_type": null, | |
| "symmetric": null | |
| } | |
| } | |
| }, | |
| "layer_type_quant_config": {}, | |
| "quant_method": "quark", | |
| "quant_mode": "eager_mode" | |
| }, | |
| "rms_norm_eps": 1e-05, | |
| "rope_theta": 1000000.0, | |
| "sliding_window": null, | |
| "tie_word_embeddings": false, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.50.3", | |
| "use_cache": true, | |
| "vocab_size": 131072 | |
| } | |