| {"_name_or_path": "lodrick-the-lafted/Densefin-291-Mistral-22B", "architectures": ["MistralForCausalLM"], "attention_dropout": 0.0, "bos_token_id": 1, "eos_token_id": 32000, "hidden_act": "silu", "hidden_size": 6144, "initializer_range": 0.02, "intermediate_size": 16384, "max_position_embeddings": 65536, "model_type": "mixtral", "num_attention_heads": 48, "num_hidden_layers": 56, "num_key_value_heads": 8, "rms_norm_eps": 1e-05, "rope_theta": 1000000, "sliding_window": null, "tie_word_embeddings": false, "torch_dtype": "bfloat16", "transformers_version": "4.40.2", "use_cache": false, "vocab_size": 32002} |