File size: 2,637 Bytes
055eba4
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
// Copyright 2024 Google LLC
// SPDX-License-Identifier: Apache-2.0
//
// Licensed under the Apache License, Version 2.0 (the "License");
// you may not use this file except in compliance with the License.
// You may obtain a copy of the License at
//
//     https://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing, software
// distributed under the License is distributed on an "AS IS" BASIS,
// WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
// See the License for the specific language governing permissions and
// limitations under the License.

#include <fstream>
#include <iostream>
#include <string>
#include <vector>

#include "compression/io.h"
#include "gemma/benchmark_helper.h"
#include "gemma/gemma.h"  // LayersOutputFunc
#include "util/args.h"
#include "hwy/base.h"
#include "nlohmann/json.hpp"

using json = nlohmann::json;

namespace gcpp {

class PromptArgs : public ArgsBase<PromptArgs> {
 public:
  PromptArgs(int argc, char* argv[]) { InitAndParse(argc, argv); }

  Path layers_output;  // optional
  std::string prompt;

  // Returns error string or nullptr if OK.
  const char* Validate() const {
    if (prompt.empty()) return "Must specify --prompt";
    return nullptr;
  }

  template <class Visitor>
  void ForEach(const Visitor& visitor) {
    visitor(layers_output, "layers_output", Path(""),
            "Path to store layers output", 2);
    visitor(prompt, "prompt", std::string(""), "Prompt to the model", 2);
  }
};

int Run(int argc, char** argv) {
  PromptArgs prompt_args(argc, argv);
  AbortIfInvalidArgs(prompt_args);

  json json_output;
  GemmaEnv env(argc, argv);
  env.MutableConfig().layers_output =
      prompt_args.layers_output.Empty()
          ? LayersOutputFunc()
          : [&json_output](int pos, const std::string& key, const float* values,
                           size_t values_len) {
              std::vector<float> v{values, values + values_len};
              json_output[std::to_string(pos)][key] = v;
            };

  const auto [answer, token_count] = env.QueryModel(prompt_args.prompt);
  std::cout << answer.substr(prompt_args.prompt.size()) << "\n" << std::flush;

  if (env.MutableConfig().layers_output) {
    std::ofstream output_f(prompt_args.layers_output.path, std::ofstream::out);
    if (!output_f) HWY_ABORT("Opening layer output file failed");
    output_f << json_output.dump();
    if (!output_f) HWY_ABORT("Writing to layer output file failed");
    output_f.close();
  }
  return 0;
}

}  // namespace gcpp

int main(int argc, char** argv) { return gcpp::Run(argc, argv); }