gpt4all/gpt4all-backend/bert_impl.h
Jared Van Bortel 4fc4d94be4
fix chat-style prompt templates (#1970)
Also use a new version of Mistral OpenOrca.

Signed-off-by: Jared Van Bortel <jared@nomic.ai>
2024-02-21 15:45:32 -05:00

46 lines
1.6 KiB
C++

#ifndef BERT_H_I_KNOW_WHAT_I_AM_DOING_WHEN_INCLUDING_THIS_FILE
#error This file is NOT meant to be included outside of bert.cpp. Doing so is DANGEROUS. Be sure to know what you are doing before proceeding to #define BERT_H_I_KNOW_WHAT_I_AM_DOING_WHEN_INCLUDING_THIS_FILE
#endif
#ifndef BERT_H
#define BERT_H
#include <string>
#include <functional>
#include <vector>
#include <memory>
#include "llmodel.h"
struct BertPrivate;
class Bert : public LLModel {
public:
Bert();
~Bert();
bool supportsEmbedding() const override { return true; }
bool supportsCompletion() const override { return true; }
bool loadModel(const std::string &modelPath, int n_ctx, int ngl) override;
bool isModelLoaded() const override;
size_t requiredMem(const std::string &modelPath, int n_ctx, int ngl) override;
size_t stateSize() const override;
size_t saveState(uint8_t *dest) const override;
size_t restoreState(const uint8_t *src) override;
void setThreadCount(int32_t n_threads) override;
int32_t threadCount() const override;
std::vector<float> embedding(const std::string &text) override;
private:
std::unique_ptr<BertPrivate> d_ptr;
protected:
std::vector<Token> tokenize(PromptContext &ctx, const std::string &str, bool special) const override;
Token sampleToken(PromptContext &ctx) const override;
std::string tokenToString(Token id) const override;
bool evalTokens(PromptContext &ctx, const std::vector<int32_t> &tokens) const override;
int32_t contextLength() const override;
const std::vector<Token> &endTokens() const override;
bool shouldAddBOS() const override { return true; }
};
#endif // BERT_H