cont : inital deslop guidelines

src/llama-ext.h

@@ -1,8 +1,14 @@
 #pragma once
 
-#include "llama-context.h"
-#include "ggml.h"
-#include "stdint.h"
+#include "llama.h"
+
+// TODO: try to remove this headers
+#include "llama-arch.h"
+#include "llama-model.h"
+#include "llama-quant.h"
+
+#include <cstdint>
+#include <vector>
 
 // Reserve a new compute graph. It is valid until the next call to llama_graph_reserve.
 LLAMA_API struct ggml_cgraph * llama_graph_reserve(
@@ -10,3 +16,29 @@ LLAMA_API struct ggml_cgraph * llama_graph_reserve(
         uint32_t n_tokens,
         uint32_t n_seqs,
         uint32_t n_outputs);
+
+LLAMA_API ggml_type llama_ftype_get_default_type(llama_ftype ftype);
+
+// TODO: use llama_quant_ prefix to name these consistently:
+
+// Returns true if this tensor should be quantized (based on name, dims, params).
+LLAMA_API bool tensor_allows_quantization(const llama_model_quantize_params * params, llm_arch arch, const ggml_tensor * tensor);
+
+// TODO: add:
+// LLAMA_API llama_quant * llama_quant_init(...);
+// LLAMA_API void llama_quant_free(llama_quant * qnt);
+
+// TODO: become member function of llama_quant
+LLAMA_API ggml_type llama_tensor_get_type(
+        llama_quant & qs,
+        const llama_model_quantize_params * params,
+        const ggml_tensor * tensor,
+        ggml_type default_type,
+        const tensor_metadata & tm);
+
+// Initialize llama_quant counters and populate tensor_metadata categories.
+// metadata: vector with name fields already set, will have category field populated.
+// TODO: become member function of llama_quant
+LLAMA_API void init_quantize_state_counters(
+        llama_quant & qs,
+        std::vector<tensor_metadata> & metadata);

src/llama-quant.cpp

@@ -2,6 +2,7 @@
 #include "llama-impl.h"
 #include "llama-model.h"
 #include "llama-model-loader.h"
+#include "llama-ext.h"
 
 #include <algorithm>
 #include <cmath>
@@ -138,7 +139,7 @@ struct compiled_tensor_type_patterns {
     std::vector<std::pair<std::regex, ggml_type>> patterns;
 };
 
-quantize_state_impl::quantize_state_impl(const llama_model & model, const llama_model_quantize_params * params)
+llama_quant::llama_quant(const llama_model & model, const llama_model_quantize_params * params)
     : model(model), params(params)
 {
     if (params->tensor_types) {
@@ -152,7 +153,7 @@ quantize_state_impl::quantize_state_impl(const llama_model & model, const llama_
     }
 }
 
-quantize_state_impl::~quantize_state_impl() = default;
+llama_quant::~llama_quant() = default;
 
 //
 // dequantization
@@ -302,7 +303,7 @@ bool tensor_allows_quantization(const llama_model_quantize_params * params, llm_
 //
 
 // incompatible tensor shapes are handled here - fallback to a compatible type
-static ggml_type tensor_type_fallback(quantize_state_impl & qs, const ggml_tensor * t, const ggml_type target_type) {
+static ggml_type tensor_type_fallback(llama_quant & qs, const ggml_tensor * t, const ggml_type target_type) {
     ggml_type return_type = target_type;
 
     const int64_t ncols = t->ne[0];
@@ -351,7 +352,7 @@ static ggml_type tensor_type_fallback(quantize_state_impl & qs, const ggml_tenso
 }
 
 // internal standard logic for selecting the target tensor type based on tensor category, ftype, and model arch
-static ggml_type llama_tensor_get_type_impl(quantize_state_impl & qs, ggml_type new_type, const ggml_tensor * tensor, llama_ftype ftype, tensor_category category) {
+static ggml_type llama_tensor_get_type_impl(llama_quant & qs, ggml_type new_type, const ggml_tensor * tensor, llama_ftype ftype, tensor_category category) {
     const std::string name = ggml_get_name(tensor);
 
     // TODO: avoid hardcoded tensor names - use the TN_* constants
@@ -601,7 +602,7 @@ static ggml_type llama_tensor_get_type_impl(quantize_state_impl & qs, ggml_type
 }
 
 // outer wrapper: determine the ggml_type that this tensor should be quantized to
-ggml_type llama_tensor_get_type(quantize_state_impl & qs, const llama_model_quantize_params * params, const ggml_tensor * tensor, ggml_type default_type, const tensor_metadata & tm) {
+ggml_type llama_tensor_get_type(llama_quant & qs, const llama_model_quantize_params * params, const ggml_tensor * tensor, ggml_type default_type, const tensor_metadata & tm) {
     if (!tensor_allows_quantization(params, qs.model.arch, tensor)) {
         return tensor->type;
     }
@@ -776,7 +777,7 @@ ggml_type llama_ftype_get_default_type(llama_ftype ftype) {
 }
 
 
-void init_quantize_state_counters(quantize_state_impl & qs, std::vector<tensor_metadata> & metadata) {
+void init_quantize_state_counters(llama_quant & qs, std::vector<tensor_metadata> & metadata) {
     for (auto & tm : metadata) {
         tensor_category cat = tensor_get_category(tm.name);
         tm.category = cat;
@@ -835,7 +836,7 @@ static void llama_model_quantize_impl(const std::string & fname_inp, const std::
     model.load_hparams(ml);
     model.load_stats  (ml);
 
-    quantize_state_impl qs(model, params);
+    llama_quant qs(model, params);
 
     if (params->only_copy) {
         ftype = ml.ftype;

src/llama-quant.h

@@ -2,16 +2,13 @@
 
 #include "llama.h"
 
-#include "ggml.h"
-
-#include "llama-arch.h"
-
 #include <memory>
 #include <string>
-#include <vector>
 
 struct llama_model;
 
+// TODO: use llama_quant_ prefix to name these consistently:
+
 // tensor categorization - used to avoid repeated string matching in quantization logic.
 // this is different from LLM_TN - we want broad categories, not specific tensor names per arch.
 enum class tensor_category {
@@ -30,6 +27,7 @@ enum class tensor_category {
 };
 
 // per-tensor metadata, computed in the preliminary loop and used in the main loop
+// TODO: probably should belong to llama_quant
 struct tensor_metadata {
     std::string     name;
     ggml_type       target_type;
@@ -48,7 +46,7 @@ struct tensor_type_option {
 
 struct compiled_tensor_type_patterns;
 
-struct quantize_state_impl {
+struct llama_quant {
     const llama_model                 & model;
     const llama_model_quantize_params * params;
 
@@ -72,16 +70,6 @@ struct quantize_state_impl {
     // tensor type override patterns (compiled once, used in llama_tensor_get_type)
     std::unique_ptr<compiled_tensor_type_patterns> tensor_type_patterns;
 
-    quantize_state_impl(const llama_model & model, const llama_model_quantize_params * params);
-    ~quantize_state_impl();
+    llama_quant(const llama_model & model, const llama_model_quantize_params * params);
+    ~llama_quant();
 };
-
-ggml_type llama_tensor_get_type(quantize_state_impl & qs, const llama_model_quantize_params * params, const ggml_tensor * tensor, ggml_type default_type, const tensor_metadata & tm);
-ggml_type llama_ftype_get_default_type(llama_ftype ftype);
-
-// Initialize quantize_state_impl counters and populate tensor_metadata categories.
-// metadata: vector with name fields already set, will have category field populated.
-void init_quantize_state_counters(quantize_state_impl & qs, std::vector<tensor_metadata> & metadata);
-
-// Returns true if this tensor should be quantized (based on name, dims, params).
-bool tensor_allows_quantization(const llama_model_quantize_params * params, llm_arch arch, const ggml_tensor * tensor);

tests/test-quant-type-selection.cpp

@@ -1,11 +1,9 @@
-#include "../src/llama-arch.h"
-#include "../src/llama-model.h"
-#include "../src/llama-quant.h"
-#include "ggml-cpp.h"
-#include "ggml.h"
-#include "gguf-model-data.h"
 #include "llama.h"
 
+#include "../src/llama-ext.h"
+
+#include "gguf-model-data.h"
+
 #include <cstdio>
 #include <cstring>
 #include <fstream>
@@ -323,13 +321,15 @@ static std::string read_file_contents(const std::string & path) {
 // ---------------------------------------------------------------------------
 
 // Returns {tensor_name, assigned_type} for each tensor, in order.
+// TODO: should likely be moved as a member function of llama_quant and expose through the `llama-ext.h` interface
 static std::vector<std::pair<std::string, ggml_type>> compute_quant_types(llama_model &                    mdl,
                                                                           const std::vector<mock_tensor> & tensors,
                                                                           llama_ftype                      ftype) {
     llama_model_quantize_params qparams = llama_model_quantize_default_params();
     qparams.ftype                       = ftype;
 
-    quantize_state_impl qs(mdl, &qparams);
+    // TODO: call llama_quant_init(...)
+    llama_quant qs(mdl, &qparams);
 
     std::vector<tensor_metadata> metadata(tensors.size());
     for (size_t i = 0; i < tensors.size(); ++i) {