mirror of
				https://github.com/ggml-org/llama.cpp.git
				synced 2025-11-03 09:22:01 +00:00 
			
		
		
		
	@@ -192,6 +192,8 @@ struct llama_hparams {
 | 
			
		||||
    // whether or not the given layer is recurrent (for hybrid models)
 | 
			
		||||
    bool is_recurrent(uint32_t il) const;
 | 
			
		||||
 | 
			
		||||
    uint32_t n_pos_per_embd() const;
 | 
			
		||||
 | 
			
		||||
    bool is_swa(uint32_t il) const;
 | 
			
		||||
};
 | 
			
		||||
 | 
			
		||||
 
 | 
			
		||||
		Reference in New Issue
	
	Block a user