| { | |
| "activation_function": "silu", | |
| "architectures": [ | |
| "ShrinkModelForCausalLM" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_shrink.ShrinkConfig", | |
| "AutoModelForCausalLM": "modeling_shrink.ShrinkModelForCausalLM" | |
| }, | |
| "bos_token_id": 1, | |
| "combined_qkv": true, | |
| "eos_token_id": 2, | |
| "hidden_size": 1024, | |
| "hidden_size_0": 8192, | |
| "initializer_range": 0.02, | |
| "intermediate_size": 4096, | |
| "layer_dropout_prob": 0.0, | |
| "layer_norm_epsilon": 1e-06, | |
| "lm_head_bias": false, | |
| "max_position_embeddings": 2048, | |
| "model_type": "shrink", | |
| "num_attention_heads": 16, | |
| "num_hidden_layers": 14, | |
| "projection_bias": true, | |
| "qk_hidden_size": null, | |
| "torch_dtype": "bfloat16", | |
| "transformers_version": "4.36.2", | |
| "use_bias": false, | |
| "use_cache": true, | |
| "vocab_size": 32000 | |
| } | |