| 1 | #include "llama.h" |
| 2 | #include "common.h" |
| 3 | |
| 4 | |
| 5 | #include <cstdio> |
| 6 | #include <cstring> |
| 7 | #include <string> |
| 8 | #include <vector> |
| 9 | #include <ctype.h> |
| 10 | #include <filesystem> |
| 11 | |
| 12 | static void print_usage(int, char ** argv) { |
| 13 | printf(format: "\nexample usage:\n" ); |
| 14 | printf(format: "\n %s -m model.gguf [-ngl n_gpu_layers] -embd-mode [-pooling] [-embd-norm <norm>] [prompt]\n" , argv[0]); |
| 15 | printf(format: "\n" ); |
| 16 | printf(format: " -embd-norm: normalization type for pooled embeddings (default: 2)\n" ); |
| 17 | printf(format: " -1=none, 0=max absolute int16, 1=taxicab, 2=Euclidean/L2, >2=p-norm\n" ); |
| 18 | printf(format: "\n" ); |
| 19 | } |
| 20 | |
| 21 | int main(int argc, char ** argv) { |
| 22 | std::string model_path; |
| 23 | std::string prompt = "Hello, my name is" ; |
| 24 | int ngl = 0; |
| 25 | bool embedding_mode = false; |
| 26 | bool pooling_enabled = false; |
| 27 | int32_t embd_norm = 2; // (-1=none, 0=max absolute int16, 1=taxicab, 2=Euclidean/L2, >2=p-norm) |
| 28 | |
| 29 | { |
| 30 | int i = 1; |
| 31 | for (; i < argc; i++) { |
| 32 | if (strcmp(s1: argv[i], s2: "-m" ) == 0) { |
| 33 | if (i + 1 < argc) { |
| 34 | model_path = argv[++i]; |
| 35 | } else { |
| 36 | print_usage(argc, argv); |
| 37 | return 1; |
| 38 | } |
| 39 | } else if (strcmp(s1: argv[i], s2: "-ngl" ) == 0) { |
| 40 | if (i + 1 < argc) { |
| 41 | try { |
| 42 | ngl = std::stoi(str: argv[++i]); |
| 43 | } catch (...) { |
| 44 | print_usage(argc, argv); |
| 45 | return 1; |
| 46 | } |
| 47 | } else { |
| 48 | print_usage(argc, argv); |
| 49 | return 1; |
| 50 | } |
| 51 | } else if (strcmp(s1: argv[i], s2: "-embd-mode" ) == 0) { |
| 52 | embedding_mode = true; |
| 53 | } else if (strcmp(s1: argv[i], s2: "-pooling" ) == 0) { |
| 54 | pooling_enabled = true; |
| 55 | } else if (strcmp(s1: argv[i], s2: "-embd-norm" ) == 0) { |
| 56 | if (i + 1 < argc) { |
| 57 | try { |
| 58 | embd_norm = std::stoi(str: argv[++i]); |
| 59 | } catch (...) { |
| 60 | print_usage(argc, argv); |
| 61 | return 1; |
| 62 | } |
| 63 | } else { |
| 64 | print_usage(argc, argv); |
| 65 | return 1; |
| 66 | } |
| 67 | } else { |
| 68 | // prompt starts here |
| 69 | break; |
| 70 | } |
| 71 | } |
| 72 | |
| 73 | if (model_path.empty()) { |
| 74 | print_usage(argc, argv); |
| 75 | return 1; |
| 76 | } |
| 77 | |
| 78 | if (i < argc) { |
| 79 | prompt = argv[i++]; |
| 80 | for (; i < argc; i++) { |
| 81 | prompt += " " ; |
| 82 | prompt += argv[i]; |
| 83 | } |
| 84 | } |
| 85 | } |
| 86 | |
| 87 | ggml_backend_load_all(); |
| 88 | llama_model_params model_params = llama_model_default_params(); |
| 89 | model_params.n_gpu_layers = ngl; |
| 90 | |
| 91 | llama_model * model = llama_model_load_from_file(path_model: model_path.c_str(), params: model_params); |
| 92 | |
| 93 | if (model == NULL) { |
| 94 | fprintf(stderr , format: "%s: error: unable to load model\n" , __func__); |
| 95 | return 1; |
| 96 | } |
| 97 | |
| 98 | // Extract basename from model_path |
| 99 | const char * basename = strrchr(s: model_path.c_str(), c: '/'); |
| 100 | basename = (basename == NULL) ? model_path.c_str() : basename + 1; |
| 101 | |
| 102 | char model_name[256]; |
| 103 | strncpy(dest: model_name, src: basename, n: 255); |
| 104 | model_name[255] = '\0'; |
| 105 | |
| 106 | char * dot = strrchr(s: model_name, c: '.'); |
| 107 | if (dot != NULL && strcmp(s1: dot, s2: ".gguf" ) == 0) { |
| 108 | *dot = '\0'; |
| 109 | } |
| 110 | printf(format: "Model name: %s\n" , model_name); |
| 111 | |
| 112 | const llama_vocab * vocab = llama_model_get_vocab(model); |
| 113 | const int n_prompt = -llama_tokenize(vocab, text: prompt.c_str(), text_len: prompt.size(), NULL, n_tokens_max: 0, add_special: true, parse_special: true); |
| 114 | |
| 115 | std::vector<llama_token> prompt_tokens(n_prompt); |
| 116 | if (llama_tokenize(vocab, text: prompt.c_str(), text_len: prompt.size(), tokens: prompt_tokens.data(), n_tokens_max: prompt_tokens.size(), add_special: true, parse_special: true) < 0) { |
| 117 | fprintf(stderr, format: "%s: error: failed to tokenize the prompt\n" , __func__); |
| 118 | return 1; |
| 119 | } |
| 120 | |
| 121 | llama_context_params ctx_params = llama_context_default_params(); |
| 122 | ctx_params.n_ctx = n_prompt; |
| 123 | ctx_params.n_batch = n_prompt; |
| 124 | ctx_params.no_perf = false; |
| 125 | if (embedding_mode) { |
| 126 | ctx_params.embeddings = true; |
| 127 | ctx_params.pooling_type = pooling_enabled ? LLAMA_POOLING_TYPE_MEAN : LLAMA_POOLING_TYPE_NONE; |
| 128 | ctx_params.n_ubatch = ctx_params.n_batch; |
| 129 | } |
| 130 | |
| 131 | llama_context * ctx = llama_init_from_model(model, params: ctx_params); |
| 132 | if (ctx == NULL) { |
| 133 | fprintf(stderr , format: "%s: error: failed to create the llama_context\n" , __func__); |
| 134 | return 1; |
| 135 | } |
| 136 | |
| 137 | printf(format: "Input prompt: \"%s\"\n" , prompt.c_str()); |
| 138 | printf(format: "Tokenized prompt (%d tokens): " , n_prompt); |
| 139 | for (auto id : prompt_tokens) { |
| 140 | char buf[128]; |
| 141 | int n = llama_token_to_piece(vocab, token: id, buf, length: sizeof(buf), lstrip: 0, special: true); |
| 142 | if (n < 0) { |
| 143 | fprintf(stderr, format: "%s: error: failed to convert token to piece\n" , __func__); |
| 144 | return 1; |
| 145 | } |
| 146 | std::string s(buf, n); |
| 147 | printf(format: "%s" , s.c_str()); |
| 148 | } |
| 149 | printf(format: "\n" ); |
| 150 | |
| 151 | llama_batch batch = llama_batch_get_one(tokens: prompt_tokens.data(), n_tokens: prompt_tokens.size()); |
| 152 | |
| 153 | if (llama_decode(ctx, batch)) { |
| 154 | fprintf(stderr, format: "%s : failed to eval\n" , __func__); |
| 155 | return 1; |
| 156 | } |
| 157 | |
| 158 | float * data_ptr; |
| 159 | int data_size; |
| 160 | const char * type; |
| 161 | std::vector<float> embd_out; |
| 162 | |
| 163 | if (embedding_mode) { |
| 164 | const int n_embd = llama_model_n_embd(model); |
| 165 | const int n_embd_count = pooling_enabled ? 1 : batch.n_tokens; |
| 166 | const int n_embeddings = n_embd * n_embd_count; |
| 167 | float * embeddings; |
| 168 | type = "-embeddings" ; |
| 169 | |
| 170 | if (llama_pooling_type(ctx) != LLAMA_POOLING_TYPE_NONE) { |
| 171 | embeddings = llama_get_embeddings_seq(ctx, seq_id: 0); |
| 172 | embd_out.resize(new_size: n_embeddings); |
| 173 | printf(format: "Normalizing embeddings using norm: %d\n" , embd_norm); |
| 174 | common_embd_normalize(inp: embeddings, out: embd_out.data(), n: n_embeddings, embd_norm); |
| 175 | embeddings = embd_out.data(); |
| 176 | } else { |
| 177 | embeddings = llama_get_embeddings(ctx); |
| 178 | } |
| 179 | |
| 180 | printf(format: "Embedding dimension: %d\n" , n_embd); |
| 181 | printf(format: "\n" ); |
| 182 | |
| 183 | // Print embeddings in the specified format |
| 184 | for (int j = 0; j < n_embd_count; j++) { |
| 185 | printf(format: "embedding %d: " , j); |
| 186 | |
| 187 | // Print first 3 values |
| 188 | for (int i = 0; i < 3 && i < n_embd; i++) { |
| 189 | printf(format: "%9.6f " , embeddings[j * n_embd + i]); |
| 190 | } |
| 191 | |
| 192 | printf(format: " ... " ); |
| 193 | |
| 194 | // Print last 3 values |
| 195 | for (int i = n_embd - 3; i < n_embd; i++) { |
| 196 | if (i >= 0) { |
| 197 | printf(format: "%9.6f " , embeddings[j * n_embd + i]); |
| 198 | } |
| 199 | } |
| 200 | |
| 201 | printf(format: "\n" ); |
| 202 | } |
| 203 | printf(format: "\n" ); |
| 204 | |
| 205 | printf(format: "Embeddings size: %d\n" , n_embeddings); |
| 206 | |
| 207 | data_ptr = embeddings; |
| 208 | data_size = n_embeddings; |
| 209 | } else { |
| 210 | float * logits = llama_get_logits_ith(ctx, i: batch.n_tokens - 1); |
| 211 | const int n_logits = llama_vocab_n_tokens(vocab); |
| 212 | type = "" ; |
| 213 | printf(format: "Vocab size: %d\n" , n_logits); |
| 214 | |
| 215 | data_ptr = logits; |
| 216 | data_size = n_logits; |
| 217 | } |
| 218 | |
| 219 | std::filesystem::create_directory(p: "data" ); |
| 220 | |
| 221 | // Save data to binary file |
| 222 | char bin_filename[512]; |
| 223 | snprintf(s: bin_filename, maxlen: sizeof(bin_filename), format: "data/llamacpp-%s%s.bin" , model_name, type); |
| 224 | printf(format: "Saving data to %s\n" , bin_filename); |
| 225 | |
| 226 | FILE * f = fopen(filename: bin_filename, modes: "wb" ); |
| 227 | if (f == NULL) { |
| 228 | fprintf(stderr, format: "%s: error: failed to open binary output file\n" , __func__); |
| 229 | return 1; |
| 230 | } |
| 231 | fwrite(ptr: data_ptr, size: sizeof(float), n: data_size, s: f); |
| 232 | fclose(stream: f); |
| 233 | |
| 234 | // Also save as text for debugging |
| 235 | char txt_filename[512]; |
| 236 | snprintf(s: txt_filename, maxlen: sizeof(txt_filename), format: "data/llamacpp-%s%s.txt" , model_name, type); |
| 237 | f = fopen(filename: txt_filename, modes: "w" ); |
| 238 | if (f == NULL) { |
| 239 | fprintf(stderr, format: "%s: error: failed to open text output file\n" , __func__); |
| 240 | return 1; |
| 241 | } |
| 242 | for (int i = 0; i < data_size; i++) { |
| 243 | fprintf(stream: f, format: "%d: %.6f\n" , i, data_ptr[i]); |
| 244 | } |
| 245 | fclose(stream: f); |
| 246 | |
| 247 | if (!embedding_mode) { |
| 248 | printf(format: "First 10 logits: " ); |
| 249 | for (int i = 0; i < 10 && i < data_size; i++) { |
| 250 | printf(format: "%.6f " , data_ptr[i]); |
| 251 | } |
| 252 | printf(format: "\n" ); |
| 253 | |
| 254 | printf(format: "Last 10 logits: " ); |
| 255 | for (int i = data_size - 10; i < data_size; i++) { |
| 256 | if (i >= 0) printf(format: "%.6f " , data_ptr[i]); |
| 257 | } |
| 258 | printf(format: "\n\n" ); |
| 259 | } |
| 260 | |
| 261 | printf(format: "Data saved to %s\n" , bin_filename); |
| 262 | printf(format: "Data saved to %s\n" , txt_filename); |
| 263 | |
| 264 | llama_free(ctx); |
| 265 | llama_model_free(model); |
| 266 | |
| 267 | return 0; |
| 268 | } |
| 269 | |