Adds server parameters for draft model cache type. Fixes ggml-org/llama.cpp/#11200

Author: aa956

common/arg.cpp

@@ -3181,6 +3181,32 @@ common_params_context common_params_parser_init(common_params & params, llama_ex
             params.speculative.model.path = value;
         }
     ).set_examples({LLAMA_EXAMPLE_SPECULATIVE, LLAMA_EXAMPLE_SERVER}).set_env("LLAMA_ARG_MODEL_DRAFT"));
+    add_opt(common_arg(
+        {"-ctkd", "--cache-type-k-draft"}, "TYPE",
+        string_format(
+            "KV cache data type for K for the draft model\n"
+            "allowed values: %s\n"
+            "(default: %s)",
+            get_all_kv_cache_types().c_str(),
+            ggml_type_name(params.speculative.cache_type_k)
+        ),
+        [](common_params & params, const std::string & value) {
+            params.speculative.cache_type_k = kv_cache_type_from_str(value);
+        }
+    ).set_env("LLAMA_ARG_CACHE_TYPE_K_DRAFT"));
+    add_opt(common_arg(
+        {"-ctvd", "--cache-type-v-draft"}, "TYPE",
+        string_format(
+            "KV cache data type for V for the draft model\n"
+            "allowed values: %s\n"
+            "(default: %s)",
+            get_all_kv_cache_types().c_str(),
+            ggml_type_name(params.speculative.cache_type_v)
+        ),
+        [](common_params & params, const std::string & value) {
+            params.speculative.cache_type_v = kv_cache_type_from_str(value);
+        }
+    ).set_env("LLAMA_ARG_CACHE_TYPE_V_DRAFT"));
 
     add_opt(common_arg(
         {"-mv", "--model-vocoder"}, "FNAME",

common/common.h

@@ -199,6 +199,9 @@ struct common_params_speculative {
     float   p_split      =  0.1f; // speculative decoding split probability
     float   p_min        = 0.75f; // minimum speculative decoding probability (greedy)
 
+    ggml_type cache_type_k = GGML_TYPE_F16; // KV cache data type for the K
+    ggml_type cache_type_v = GGML_TYPE_F16; // KV cache data type for the V
+
     struct cpu_params cpuparams;
     struct cpu_params cpuparams_batch;
 

tools/server/README.md

@@ -186,6 +186,8 @@ The project is under active development, and we are [looking for feedback and co
 | `-devd, --device-draft <dev1,dev2,..>` | comma-separated list of devices to use for offloading the draft model (none = don't offload)<br/>use --list-devices to see a list of available devices |
 | `-ngld, --gpu-layers-draft, --n-gpu-layers-draft N` | number of layers to store in VRAM for the draft model<br/>(env: LLAMA_ARG_N_GPU_LAYERS_DRAFT) |
 | `-md, --model-draft FNAME` | draft model for speculative decoding (default: unused)<br/>(env: LLAMA_ARG_MODEL_DRAFT) |
+| `-ctkd, --cache-type-k-draft TYPE` | KV cache data type for K for speculative decoding model<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_K_DRAFT) |
+| `-ctvd, --cache-type-v-draft TYPE` | KV cache data type for V for speculative decoding model<br/>allowed values: f32, f16, bf16, q8_0, q4_0, q4_1, iq4_nl, q5_0, q5_1<br/>(default: f16)<br/>(env: LLAMA_ARG_CACHE_TYPE_V_DRAFT) |
 | `-mv, --model-vocoder FNAME` | vocoder model for audio generation (default: unused) |
 | `--tts-use-guide-tokens` | Use guide tokens to improve TTS word recall |
 | `--embd-bge-small-en-default` | use default bge-small-en-v1.5 model (note: can download weights from the internet) |

tools/server/server.cpp

@@ -1939,10 +1939,8 @@ struct server_context {
             params_dft.n_ctx        = params_base.speculative.n_ctx == 0 ? params_base.n_ctx / params_base.n_parallel : params_base.speculative.n_ctx;
             params_dft.n_gpu_layers = params_base.speculative.n_gpu_layers;
             params_dft.n_parallel   = 1;
-
-            // force F16 KV cache for the draft model for extra performance
-            params_dft.cache_type_k = GGML_TYPE_F16;
-            params_dft.cache_type_v = GGML_TYPE_F16;
+            params_dft.cache_type_k = params_base.speculative.cache_type_k;
+            params_dft.cache_type_v = params_base.speculative.cache_type_v;
 
             llama_init_dft = common_init_from_params(params_dft);