github/gpt4all
1
0
Fork 0
mirror of https://github.com/nomic-ai/gpt4all.git synced 2025-09-06 11:00:48 +00:00
Code Issues Packages Projects Releases Wiki Activity

Dlopen backend 5 (#779)

Browse Source 
Major change to the backend that allows for pluggable versions of llama.cpp/ggml. This was squashed merged from dlopen_backend_5 where the history is preserved.
This commit is contained in:
AT
2023-05-31 17:04:01 -04:00
committed by GitHub
parent f4a1f7340c
commit 48275d0dcc
22 changed files with 993 additions and 327 deletions
Download Patch File Download Diff File Expand all files Collapse all files

16
gpt4all-backend/llmodel.h
View File

@@ -1,21 +1,23 @@
#ifndef LLMODEL_H
#define LLMODEL_H
#include <string>
#include <functional>
#include <vector>
#include <cstdint>
class LLModel {
public:
explicit LLModel() {}
virtual ~LLModel() {}
static LLModel *construct(const std::string &modelPath, std::string buildVariant = "default");
virtual bool loadModel(const std::string &modelPath) = 0;
virtual bool isModelLoaded() const = 0;
virtual size_t stateSize() const { return 0; }
virtual size_t saveState(uint8_t *dest) const { return 0; }
virtual size_t restoreState(const uint8_t *src) { return 0; }
virtual size_t saveState(uint8_t */*dest*/) const { return 0; }
virtual size_t restoreState(const uint8_t */*src*/) { return 0; }
struct PromptContext {
std::vector<float> logits; // logits of current context
std::vector<int32_t> tokens; // current tokens in the context window
@@ -36,12 +38,18 @@ public:
std::function<bool(int32_t, const std::string&)> responseCallback,
std::function<bool(bool)> recalculateCallback,
PromptContext &ctx) = 0;
virtual void setThreadCount(int32_t n_threads) {}
virtual void setThreadCount(int32_t /*n_threads*/) {}
virtual int32_t threadCount() const { return 1; }
const char *getModelType() const {
return modelType;
}
protected:
virtual void recalculateContext(PromptContext &promptCtx,
std::function<bool(bool)> recalculate) = 0;
const char *modelType;
};
#endif // LLMODEL_H