{ "_name_or_path": "huggingface_repos/Llama-2-13b-hf-asvd85", "architectures": [ "ASVDLlamaForCausalLM" ], "attention_bias": false, "auto_map": { "AutoConfig": "configuration_asvd_llama.ASVDLlamaConfig", "AutoModelForCausalLM": "modeling_asvd_llama.ASVDLlamaForCausalLM" }, "bos_token_id": 1, "eos_token_id": 2, "hidden_act": "silu", "hidden_size": 5120, "initializer_range": 0.02, "intermediate_size": 13824, "max_position_embeddings": 4096, "model_type": "llama", "num_attention_heads": 40, "num_hidden_layers": 40, "num_key_value_heads": 40, "pretraining_tp": 1, "rms_norm_eps": 1e-05, "rope_scaling": null, "rope_theta": 10000.0, "tie_word_embeddings": false, "torch_dtype": "float16", "transformers_version": "4.35.2", "truncation_ranks": { "model.layers.0.mlp.down_proj": 2241, "model.layers.0.mlp.gate_proj": 1494, "model.layers.0.mlp.up_proj": 1494, "model.layers.0.self_attn.k_proj": 256, "model.layers.0.self_attn.o_proj": 256, "model.layers.0.self_attn.q_proj": 256, "model.layers.0.self_attn.v_proj": 256, "model.layers.1.mlp.gate_proj": 1868, "model.layers.1.mlp.up_proj": 2615, "model.layers.1.self_attn.k_proj": 256, "model.layers.1.self_attn.o_proj": 1280, "model.layers.1.self_attn.q_proj": 512, "model.layers.10.self_attn.k_proj": 1024, "model.layers.10.self_attn.o_proj": 2304, "model.layers.10.self_attn.q_proj": 1280, "model.layers.11.mlp.gate_proj": 2988, "model.layers.11.self_attn.k_proj": 1536, "model.layers.11.self_attn.q_proj": 1280, "model.layers.12.self_attn.k_proj": 1536, "model.layers.12.self_attn.q_proj": 1280, "model.layers.13.mlp.up_proj": 3362, "model.layers.13.self_attn.k_proj": 1024, "model.layers.13.self_attn.q_proj": 1024, "model.layers.14.mlp.down_proj": 3362, "model.layers.14.mlp.gate_proj": 3362, "model.layers.14.self_attn.k_proj": 1536, "model.layers.14.self_attn.o_proj": 2304, "model.layers.14.self_attn.q_proj": 1024, "model.layers.15.self_attn.k_proj": 1024, "model.layers.15.self_attn.q_proj": 1280, "model.layers.16.self_attn.k_proj": 1024, "model.layers.16.self_attn.q_proj": 768, "model.layers.17.self_attn.k_proj": 1536, "model.layers.17.self_attn.o_proj": 2048, "model.layers.17.self_attn.q_proj": 1280, "model.layers.17.self_attn.v_proj": 2304, "model.layers.18.mlp.gate_proj": 2988, "model.layers.18.self_attn.k_proj": 768, "model.layers.18.self_attn.o_proj": 2304, "model.layers.18.self_attn.q_proj": 1024, "model.layers.18.self_attn.v_proj": 2048, "model.layers.19.self_attn.k_proj": 1536, "model.layers.19.self_attn.o_proj": 2048, "model.layers.19.self_attn.q_proj": 1280, "model.layers.2.mlp.gate_proj": 3362, "model.layers.2.mlp.up_proj": 3362, "model.layers.2.self_attn.k_proj": 768, "model.layers.2.self_attn.o_proj": 2048, "model.layers.2.self_attn.q_proj": 1024, "model.layers.20.mlp.down_proj": 3362, "model.layers.20.mlp.up_proj": 3362, "model.layers.20.self_attn.k_proj": 1280, "model.layers.20.self_attn.o_proj": 2304, "model.layers.20.self_attn.q_proj": 1792, "model.layers.21.self_attn.k_proj": 1024, "model.layers.21.self_attn.o_proj": 2304, "model.layers.21.self_attn.q_proj": 1024, "model.layers.22.self_attn.k_proj": 1024, "model.layers.22.self_attn.o_proj": 1792, "model.layers.22.self_attn.q_proj": 1024, "model.layers.22.self_attn.v_proj": 2048, "model.layers.23.self_attn.k_proj": 256, "model.layers.23.self_attn.o_proj": 1792, "model.layers.23.self_attn.q_proj": 256, "model.layers.23.self_attn.v_proj": 2304, "model.layers.24.self_attn.k_proj": 1536, "model.layers.24.self_attn.o_proj": 1792, "model.layers.24.self_attn.q_proj": 2048, "model.layers.24.self_attn.v_proj": 2048, "model.layers.25.mlp.up_proj": 3362, "model.layers.25.self_attn.k_proj": 768, "model.layers.25.self_attn.o_proj": 1536, "model.layers.25.self_attn.q_proj": 1024, "model.layers.25.self_attn.v_proj": 2048, "model.layers.26.self_attn.k_proj": 768, "model.layers.26.self_attn.o_proj": 1536, "model.layers.26.self_attn.q_proj": 1024, "model.layers.26.self_attn.v_proj": 1792, "model.layers.27.self_attn.k_proj": 1280, "model.layers.27.self_attn.o_proj": 2304, "model.layers.27.self_attn.q_proj": 1280, "model.layers.27.self_attn.v_proj": 2048, "model.layers.28.self_attn.k_proj": 1280, "model.layers.28.self_attn.o_proj": 1536, "model.layers.28.self_attn.q_proj": 1280, "model.layers.28.self_attn.v_proj": 1792, "model.layers.29.self_attn.k_proj": 1536, "model.layers.29.self_attn.q_proj": 1536, "model.layers.29.self_attn.v_proj": 2304, "model.layers.3.mlp.down_proj": 2988, "model.layers.3.self_attn.k_proj": 768, "model.layers.3.self_attn.o_proj": 1024, "model.layers.3.self_attn.q_proj": 512, "model.layers.30.self_attn.k_proj": 1536, "model.layers.30.self_attn.o_proj": 1792, "model.layers.30.self_attn.q_proj": 2048, "model.layers.30.self_attn.v_proj": 2048, "model.layers.31.self_attn.k_proj": 768, "model.layers.31.self_attn.o_proj": 1792, "model.layers.31.self_attn.q_proj": 1024, "model.layers.31.self_attn.v_proj": 1024, "model.layers.32.self_attn.k_proj": 1280, "model.layers.32.self_attn.o_proj": 2048, "model.layers.32.self_attn.q_proj": 1280, "model.layers.32.self_attn.v_proj": 2304, "model.layers.33.mlp.gate_proj": 3362, "model.layers.33.self_attn.k_proj": 1024, "model.layers.33.self_attn.o_proj": 1280, "model.layers.33.self_attn.q_proj": 1280, "model.layers.34.self_attn.k_proj": 1280, "model.layers.34.self_attn.o_proj": 1024, "model.layers.34.self_attn.q_proj": 1536, "model.layers.34.self_attn.v_proj": 1024, "model.layers.35.self_attn.k_proj": 1024, "model.layers.35.self_attn.o_proj": 1280, "model.layers.35.self_attn.q_proj": 1280, "model.layers.35.self_attn.v_proj": 2048, "model.layers.36.self_attn.k_proj": 1280, "model.layers.36.self_attn.o_proj": 1280, "model.layers.36.self_attn.q_proj": 768, "model.layers.36.self_attn.v_proj": 2304, "model.layers.37.self_attn.k_proj": 1024, "model.layers.37.self_attn.o_proj": 1536, "model.layers.37.self_attn.q_proj": 1024, "model.layers.38.self_attn.k_proj": 2048, "model.layers.38.self_attn.o_proj": 1792, "model.layers.38.self_attn.q_proj": 1024, "model.layers.39.self_attn.k_proj": 512, "model.layers.39.self_attn.o_proj": 1280, "model.layers.39.self_attn.q_proj": 768, "model.layers.4.mlp.down_proj": 2988, "model.layers.4.self_attn.k_proj": 1024, "model.layers.4.self_attn.o_proj": 2304, "model.layers.4.self_attn.q_proj": 1536, "model.layers.5.mlp.gate_proj": 3362, "model.layers.5.self_attn.k_proj": 768, "model.layers.5.self_attn.o_proj": 1280, "model.layers.5.self_attn.q_proj": 512, "model.layers.6.self_attn.k_proj": 1024, "model.layers.6.self_attn.o_proj": 1792, "model.layers.6.self_attn.q_proj": 1024, "model.layers.7.self_attn.k_proj": 1280, "model.layers.7.self_attn.o_proj": 2304, "model.layers.7.self_attn.q_proj": 1024, "model.layers.7.self_attn.v_proj": 1536, "model.layers.8.self_attn.k_proj": 768, "model.layers.8.self_attn.o_proj": 1792, "model.layers.8.self_attn.q_proj": 1024, "model.layers.8.self_attn.v_proj": 2304, "model.layers.9.self_attn.k_proj": 512, "model.layers.9.self_attn.o_proj": 2048, "model.layers.9.self_attn.q_proj": 1024 }, "use_cache": true, "vocab_size": 32000 }