| { | |
| "_name_or_path": "huggingface_repos/Llama-2-7b-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": 4096, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 11008, | |
| "max_position_embeddings": 4096, | |
| "model_type": "llama", | |
| "num_attention_heads": 32, | |
| "num_hidden_layers": 32, | |
| "num_key_value_heads": 32, | |
| "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": 1791, | |
| "model.layers.0.mlp.gate_proj": 1194, | |
| "model.layers.0.mlp.up_proj": 1194, | |
| "model.layers.0.self_attn.k_proj": 204, | |
| "model.layers.0.self_attn.o_proj": 204, | |
| "model.layers.0.self_attn.q_proj": 204, | |
| "model.layers.0.self_attn.v_proj": 614, | |
| "model.layers.1.mlp.gate_proj": 1791, | |
| "model.layers.1.mlp.up_proj": 2089, | |
| "model.layers.1.self_attn.k_proj": 204, | |
| "model.layers.1.self_attn.o_proj": 819, | |
| "model.layers.1.self_attn.q_proj": 204, | |
| "model.layers.1.self_attn.v_proj": 1024, | |
| "model.layers.10.self_attn.k_proj": 1228, | |
| "model.layers.10.self_attn.q_proj": 1024, | |
| "model.layers.11.self_attn.k_proj": 409, | |
| "model.layers.11.self_attn.o_proj": 1638, | |
| "model.layers.11.self_attn.q_proj": 409, | |
| "model.layers.12.mlp.down_proj": 2686, | |
| "model.layers.12.mlp.gate_proj": 1492, | |
| "model.layers.12.mlp.up_proj": 1791, | |
| "model.layers.12.self_attn.k_proj": 1024, | |
| "model.layers.12.self_attn.o_proj": 1433, | |
| "model.layers.12.self_attn.q_proj": 614, | |
| "model.layers.13.self_attn.k_proj": 1433, | |
| "model.layers.13.self_attn.o_proj": 1638, | |
| "model.layers.13.self_attn.q_proj": 1228, | |
| "model.layers.14.self_attn.k_proj": 1228, | |
| "model.layers.14.self_attn.q_proj": 1024, | |
| "model.layers.15.self_attn.k_proj": 1433, | |
| "model.layers.15.self_attn.q_proj": 1024, | |
| "model.layers.16.self_attn.k_proj": 614, | |
| "model.layers.16.self_attn.o_proj": 1843, | |
| "model.layers.16.self_attn.q_proj": 819, | |
| "model.layers.17.self_attn.k_proj": 1024, | |
| "model.layers.17.self_attn.q_proj": 1228, | |
| "model.layers.18.self_attn.k_proj": 1433, | |
| "model.layers.18.self_attn.o_proj": 1638, | |
| "model.layers.18.self_attn.q_proj": 1024, | |
| "model.layers.19.self_attn.k_proj": 1228, | |
| "model.layers.19.self_attn.o_proj": 1843, | |
| "model.layers.19.self_attn.q_proj": 1433, | |
| "model.layers.2.mlp.gate_proj": 2686, | |
| "model.layers.2.self_attn.k_proj": 819, | |
| "model.layers.2.self_attn.o_proj": 1638, | |
| "model.layers.2.self_attn.q_proj": 1024, | |
| "model.layers.2.self_attn.v_proj": 1843, | |
| "model.layers.20.self_attn.k_proj": 204, | |
| "model.layers.20.self_attn.o_proj": 1433, | |
| "model.layers.20.self_attn.q_proj": 409, | |
| "model.layers.20.self_attn.v_proj": 1638, | |
| "model.layers.21.self_attn.k_proj": 1024, | |
| "model.layers.21.self_attn.o_proj": 1843, | |
| "model.layers.21.self_attn.q_proj": 1228, | |
| "model.layers.21.self_attn.v_proj": 1433, | |
| "model.layers.22.self_attn.k_proj": 1228, | |
| "model.layers.22.self_attn.o_proj": 1638, | |
| "model.layers.22.self_attn.q_proj": 1433, | |
| "model.layers.23.mlp.up_proj": 2686, | |
| "model.layers.23.self_attn.k_proj": 1024, | |
| "model.layers.23.self_attn.o_proj": 1638, | |
| "model.layers.23.self_attn.q_proj": 1024, | |
| "model.layers.23.self_attn.v_proj": 1638, | |
| "model.layers.24.mlp.down_proj": 2686, | |
| "model.layers.24.mlp.gate_proj": 2089, | |
| "model.layers.24.mlp.up_proj": 2089, | |
| "model.layers.24.self_attn.k_proj": 204, | |
| "model.layers.24.self_attn.o_proj": 819, | |
| "model.layers.24.self_attn.q_proj": 204, | |
| "model.layers.24.self_attn.v_proj": 1433, | |
| "model.layers.25.mlp.gate_proj": 2686, | |
| "model.layers.25.mlp.up_proj": 2686, | |
| "model.layers.25.self_attn.o_proj": 1433, | |
| "model.layers.25.self_attn.q_proj": 1638, | |
| "model.layers.26.self_attn.k_proj": 819, | |
| "model.layers.26.self_attn.o_proj": 1228, | |
| "model.layers.26.self_attn.q_proj": 1024, | |
| "model.layers.27.self_attn.k_proj": 614, | |
| "model.layers.27.self_attn.o_proj": 1638, | |
| "model.layers.27.self_attn.q_proj": 819, | |
| "model.layers.27.self_attn.v_proj": 1843, | |
| "model.layers.28.mlp.gate_proj": 2686, | |
| "model.layers.28.self_attn.k_proj": 1433, | |
| "model.layers.28.self_attn.o_proj": 1433, | |
| "model.layers.28.self_attn.q_proj": 1228, | |
| "model.layers.29.self_attn.k_proj": 1228, | |
| "model.layers.29.self_attn.o_proj": 1433, | |
| "model.layers.29.self_attn.q_proj": 819, | |
| "model.layers.3.mlp.down_proj": 2686, | |
| "model.layers.3.mlp.gate_proj": 2686, | |
| "model.layers.3.self_attn.k_proj": 409, | |
| "model.layers.3.self_attn.o_proj": 1024, | |
| "model.layers.3.self_attn.q_proj": 819, | |
| "model.layers.30.self_attn.k_proj": 1638, | |
| "model.layers.30.self_attn.o_proj": 1638, | |
| "model.layers.30.self_attn.q_proj": 1433, | |
| "model.layers.31.self_attn.k_proj": 1024, | |
| "model.layers.31.self_attn.o_proj": 1228, | |
| "model.layers.31.self_attn.q_proj": 1024, | |
| "model.layers.4.mlp.down_proj": 2686, | |
| "model.layers.4.mlp.gate_proj": 2686, | |
| "model.layers.4.self_attn.k_proj": 614, | |
| "model.layers.4.self_attn.o_proj": 1433, | |
| "model.layers.4.self_attn.q_proj": 819, | |
| "model.layers.5.self_attn.k_proj": 1433, | |
| "model.layers.5.self_attn.o_proj": 1024, | |
| "model.layers.5.self_attn.q_proj": 409, | |
| "model.layers.6.self_attn.k_proj": 614, | |
| "model.layers.6.self_attn.o_proj": 1638, | |
| "model.layers.6.self_attn.q_proj": 614, | |
| "model.layers.7.mlp.gate_proj": 2686, | |
| "model.layers.7.self_attn.k_proj": 819, | |
| "model.layers.7.self_attn.o_proj": 1433, | |
| "model.layers.7.self_attn.q_proj": 1024, | |
| "model.layers.8.self_attn.k_proj": 1024, | |
| "model.layers.8.self_attn.o_proj": 1433, | |
| "model.layers.8.self_attn.q_proj": 1024, | |
| "model.layers.9.self_attn.k_proj": 1024, | |
| "model.layers.9.self_attn.o_proj": 1843, | |
| "model.layers.9.self_attn.q_proj": 1024 | |
| }, | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |