Skip to content

Commit b16fb0d

Browse files
l3utterflyl3utterfly
committed
removed unused functions
1 parent 3e9852b commit b16fb0d

File tree

2 files changed

+1
-224
lines changed

2 files changed

+1
-224
lines changed

‎common/common.cpp‎

Lines changed: 0 additions & 211 deletions
Original file line numberDiff line numberDiff line change
@@ -1952,214 +1952,3 @@ common_control_vector_data common_control_vector_load(const std::vector<common_c
19521952

19531953
return result;
19541954
}
1955-
1956-
//
1957-
// YAML utils
1958-
//
1959-
1960-
void yaml_dump_vector_float(FILE * stream, const char * prop_name, const std::vector<float> & data) {
1961-
if (data.empty()) {
1962-
fprintf(stream, "%s:\n", prop_name);
1963-
return;
1964-
}
1965-
1966-
fprintf(stream, "%s: [", prop_name);
1967-
for (size_t i = 0; i < data.size() - 1; ++i) {
1968-
fprintf(stream, "%e, ", data[i]);
1969-
}
1970-
fprintf(stream, "%e]\n", data.back());
1971-
}
1972-
1973-
void yaml_dump_vector_int(FILE * stream, const char * prop_name, const std::vector<int> & data) {
1974-
if (data.empty()) {
1975-
fprintf(stream, "%s:\n", prop_name);
1976-
return;
1977-
}
1978-
1979-
fprintf(stream, "%s: [", prop_name);
1980-
for (size_t i = 0; i < data.size() - 1; ++i) {
1981-
fprintf(stream, "%d, ", data[i]);
1982-
}
1983-
fprintf(stream, "%d]\n", data.back());
1984-
}
1985-
1986-
void yaml_dump_string_multiline(FILE * stream, const char * prop_name, const char * data) {
1987-
std::string data_str(data == NULL ? "" : data);
1988-
1989-
if (data_str.empty()) {
1990-
fprintf(stream, "%s:\n", prop_name);
1991-
return;
1992-
}
1993-
1994-
size_t pos_start = 0;
1995-
size_t pos_found = 0;
1996-
1997-
if (std::isspace(data_str[0]) || std::isspace(data_str.back())) {
1998-
data_str = std::regex_replace(data_str, std::regex("\n"), "\\n");
1999-
data_str = std::regex_replace(data_str, std::regex("\""), "\\\"");
2000-
data_str = std::regex_replace(data_str, std::regex(R"(\\[^n"])"), R"(\$&)");
2001-
data_str = "\"" + data_str + "\"";
2002-
fprintf(stream, "%s: %s\n", prop_name, data_str.c_str());
2003-
return;
2004-
}
2005-
2006-
if (data_str.find('\n') == std::string::npos) {
2007-
fprintf(stream, "%s: %s\n", prop_name, data_str.c_str());
2008-
return;
2009-
}
2010-
2011-
fprintf(stream, "%s: |\n", prop_name);
2012-
while ((pos_found = data_str.find('\n', pos_start)) != std::string::npos) {
2013-
fprintf(stream, " %s\n", data_str.substr(pos_start, pos_found-pos_start).c_str());
2014-
pos_start = pos_found + 1;
2015-
}
2016-
}
2017-
2018-
void yaml_dump_non_result_info(FILE * stream, const common_params & params, const llama_context * lctx,
2019-
const std::string & timestamp, const std::vector<int> & prompt_tokens, const char * model_desc) {
2020-
ggml_cpu_init(); // some ARM features are detected at runtime
2021-
2022-
const auto & sparams = params.sampling;
2023-
2024-
fprintf(stream, "build_commit: %s\n", LLAMA_COMMIT);
2025-
fprintf(stream, "build_number: %d\n", LLAMA_BUILD_NUMBER);
2026-
fprintf(stream, "cpu_has_arm_fma: %s\n", ggml_cpu_has_arm_fma() ? "true" : "false");
2027-
fprintf(stream, "cpu_has_avx: %s\n", ggml_cpu_has_avx() ? "true" : "false");
2028-
fprintf(stream, "cpu_has_avx_vnni: %s\n", ggml_cpu_has_avx_vnni() ? "true" : "false");
2029-
fprintf(stream, "cpu_has_avx2: %s\n", ggml_cpu_has_avx2() ? "true" : "false");
2030-
fprintf(stream, "cpu_has_avx512: %s\n", ggml_cpu_has_avx512() ? "true" : "false");
2031-
fprintf(stream, "cpu_has_avx512_vbmi: %s\n", ggml_cpu_has_avx512_vbmi() ? "true" : "false");
2032-
fprintf(stream, "cpu_has_avx512_vnni: %s\n", ggml_cpu_has_avx512_vnni() ? "true" : "false");
2033-
fprintf(stream, "cpu_has_fma: %s\n", ggml_cpu_has_fma() ? "true" : "false");
2034-
fprintf(stream, "cpu_has_neon: %s\n", ggml_cpu_has_neon() ? "true" : "false");
2035-
fprintf(stream, "cpu_has_sve: %s\n", ggml_cpu_has_sve() ? "true" : "false");
2036-
fprintf(stream, "cpu_has_f16c: %s\n", ggml_cpu_has_f16c() ? "true" : "false");
2037-
fprintf(stream, "cpu_has_fp16_va: %s\n", ggml_cpu_has_fp16_va() ? "true" : "false");
2038-
fprintf(stream, "cpu_has_riscv_v: %s\n", ggml_cpu_has_riscv_v() ? "true" : "false");
2039-
fprintf(stream, "cpu_has_wasm_simd: %s\n", ggml_cpu_has_wasm_simd() ? "true" : "false");
2040-
fprintf(stream, "cpu_has_sse3: %s\n", ggml_cpu_has_sse3() ? "true" : "false");
2041-
fprintf(stream, "cpu_has_vsx: %s\n", ggml_cpu_has_vsx() ? "true" : "false");
2042-
fprintf(stream, "cpu_has_matmul_int8: %s\n", ggml_cpu_has_matmul_int8() ? "true" : "false");
2043-
2044-
#ifdef NDEBUG
2045-
fprintf(stream, "debug: false\n");
2046-
#else
2047-
fprintf(stream, "debug: true\n");
2048-
#endif // NDEBUG
2049-
2050-
fprintf(stream, "model_desc: %s\n", model_desc);
2051-
fprintf(stream, "n_vocab: %d # output size of the final layer, 32001 for some models\n", llama_n_vocab(llama_get_model(lctx)));
2052-
2053-
#ifdef __OPTIMIZE__
2054-
fprintf(stream, "optimize: true\n");
2055-
#else
2056-
fprintf(stream, "optimize: false\n");
2057-
#endif // __OPTIMIZE__
2058-
2059-
fprintf(stream, "time: %s\n", timestamp.c_str());
2060-
2061-
fprintf(stream, "\n");
2062-
fprintf(stream, "###############\n");
2063-
fprintf(stream, "# User Inputs #\n");
2064-
fprintf(stream, "###############\n");
2065-
fprintf(stream, "\n");
2066-
2067-
fprintf(stream, "alias: %s # default: unknown\n", params.model_alias.c_str());
2068-
fprintf(stream, "batch_size: %d # default: 512\n", params.n_batch);
2069-
fprintf(stream, "chunks: %d # default: -1 (unlimited)\n", params.n_chunks);
2070-
fprintf(stream, "color: %s # default: false\n", params.use_color ? "true" : "false");
2071-
fprintf(stream, "ctx_size: %d # default: 512\n", params.n_ctx);
2072-
fprintf(stream, "dry_allowed_length: %d # default: 2\n", sparams.dry_allowed_length);
2073-
fprintf(stream, "dry_base: %.2f # default: 1.75\n", sparams.dry_base);
2074-
fprintf(stream, "dry_multiplier: %.1f # default: 0.0\n", sparams.dry_multiplier);
2075-
fprintf(stream, "dry_penalty_last_n: %d # default: -1 (0 = disable, -1 = context size)\n", sparams.dry_penalty_last_n);
2076-
fprintf(stream, "escape: %s # default: false\n", params.escape ? "true" : "false");
2077-
fprintf(stream, "file: # never logged, see prompt instead. Can still be specified for input.\n");
2078-
fprintf(stream, "frequency_penalty: %f # default: 0.0 \n", sparams.penalty_freq);
2079-
yaml_dump_string_multiline(stream, "grammar", sparams.grammar.c_str());
2080-
fprintf(stream, "grammar-file: # never logged, see grammar instead. Can still be specified for input.\n");
2081-
fprintf(stream, "hellaswag: %s # default: false\n", params.hellaswag ? "true" : "false");
2082-
fprintf(stream, "hellaswag_tasks: %zu # default: 400\n", params.hellaswag_tasks);
2083-
fprintf(stream, "ignore_eos: %s # default: false\n", sparams.ignore_eos ? "true" : "false");
2084-
2085-
yaml_dump_string_multiline(stream, "in_prefix", params.input_prefix.c_str());
2086-
fprintf(stream, "in_prefix_bos: %s # default: false\n", params.input_prefix_bos ? "true" : "false");
2087-
yaml_dump_string_multiline(stream, "in_suffix", params.input_prefix.c_str());
2088-
fprintf(stream, "interactive: %s # default: false\n", params.interactive ? "true" : "false");
2089-
fprintf(stream, "interactive_first: %s # default: false\n", params.interactive_first ? "true" : "false");
2090-
fprintf(stream, "keep: %d # default: 0\n", params.n_keep);
2091-
//fprintf(stream, "logdir: %s # default: unset (no logging)\n", params.logdir.c_str());
2092-
2093-
fprintf(stream, "logit_bias:\n");
2094-
for (const auto & logit_bias : sparams.logit_bias) {
2095-
fprintf(stream, " %d: %f", logit_bias.token, logit_bias.bias);
2096-
}
2097-
2098-
fprintf(stream, "lora:\n");
2099-
for (auto & la : params.lora_adapters) {
2100-
if (la.scale == 1.0f) {
2101-
fprintf(stream, " - %s\n", la.path.c_str());
2102-
}
2103-
}
2104-
fprintf(stream, "lora_scaled:\n");
2105-
for (auto & la : params.lora_adapters) {
2106-
if (la.scale != 1.0f) {
2107-
fprintf(stream, " - %s: %f\n", la.path.c_str(), la.scale);
2108-
}
2109-
}
2110-
fprintf(stream, "lora_init_without_apply: %s # default: false\n", params.lora_init_without_apply ? "true" : "false");
2111-
fprintf(stream, "main_gpu: %d # default: 0\n", params.main_gpu);
2112-
fprintf(stream, "min_keep: %d # default: 0 (disabled)\n", sparams.min_keep);
2113-
fprintf(stream, "mirostat: %d # default: 0 (disabled)\n", sparams.mirostat);
2114-
fprintf(stream, "mirostat_ent: %f # default: 5.0\n", sparams.mirostat_tau);
2115-
fprintf(stream, "mirostat_lr: %f # default: 0.1\n", sparams.mirostat_eta);
2116-
fprintf(stream, "mlock: %s # default: false\n", params.use_mlock ? "true" : "false");
2117-
fprintf(stream, "model: %s # default: %s\n", params.model.c_str(), DEFAULT_MODEL_PATH);
2118-
//fprintf(stream, "model_draft: %s # default:\n", params.model_draft.c_str());
2119-
fprintf(stream, "multiline_input: %s # default: false\n", params.multiline_input ? "true" : "false");
2120-
fprintf(stream, "n_gpu_layers: %d # default: -1\n", params.n_gpu_layers);
2121-
fprintf(stream, "n_predict: %d # default: -1 (unlimited)\n", params.n_predict);
2122-
fprintf(stream, "n_probs: %d # only used by server binary, default: 0\n", sparams.n_probs);
2123-
fprintf(stream, "no_mmap: %s # default: false\n", !params.use_mmap ? "true" : "false");
2124-
fprintf(stream, "penalize_nl: %s # default: false\n", sparams.penalize_nl ? "true" : "false");
2125-
fprintf(stream, "ppl_output_type: %d # default: 0\n", params.ppl_output_type);
2126-
fprintf(stream, "ppl_stride: %d # default: 0\n", params.ppl_stride);
2127-
fprintf(stream, "presence_penalty: %f # default: 0.0\n", sparams.penalty_present);
2128-
yaml_dump_string_multiline(stream, "prompt", params.prompt.c_str());
2129-
fprintf(stream, "prompt_cache: %s\n", params.path_prompt_cache.c_str());
2130-
fprintf(stream, "prompt_cache_all: %s # default: false\n", params.prompt_cache_all ? "true" : "false");
2131-
fprintf(stream, "prompt_cache_ro: %s # default: false\n", params.prompt_cache_ro ? "true" : "false");
2132-
yaml_dump_vector_int(stream, "prompt_tokens", prompt_tokens);
2133-
fprintf(stream, "repeat_penalty: %f # default: 1.1\n", sparams.penalty_repeat);
2134-
2135-
fprintf(stream, "reverse_prompt:\n");
2136-
for (std::string ap : params.antiprompt) {
2137-
size_t pos = 0;
2138-
while ((pos = ap.find('\n', pos)) != std::string::npos) {
2139-
ap.replace(pos, 1, "\\n");
2140-
pos += 1;
2141-
}
2142-
2143-
fprintf(stream, " - %s\n", ap.c_str());
2144-
}
2145-
2146-
fprintf(stream, "rope_freq_base: %f # default: 10000.0\n", params.rope_freq_base);
2147-
fprintf(stream, "rope_freq_scale: %f # default: 1.0\n", params.rope_freq_scale);
2148-
fprintf(stream, "simple_io: %s # default: false\n", params.simple_io ? "true" : "false");
2149-
fprintf(stream, "cont_batching: %s # default: false\n", params.cont_batching ? "true" : "false");
2150-
fprintf(stream, "flash_attn: %s # default: false\n", params.flash_attn ? "true" : "false");
2151-
fprintf(stream, "temp: %f # default: 0.8\n", sparams.temp);
2152-
2153-
const std::vector<float> tensor_split_vector(params.tensor_split, params.tensor_split + llama_max_devices());
2154-
yaml_dump_vector_float(stream, "tensor_split", tensor_split_vector);
2155-
2156-
fprintf(stream, "threads: %d # default: %u\n", params.cpuparams.n_threads, std::thread::hardware_concurrency());
2157-
fprintf(stream, "top_k: %d # default: 40\n", sparams.top_k);
2158-
fprintf(stream, "top_p: %f # default: 0.95\n", sparams.top_p);
2159-
fprintf(stream, "min_p: %f # default: 0.0\n", sparams.min_p);
2160-
fprintf(stream, "xtc_probability: %f # default: 0.0\n", sparams.xtc_probability);
2161-
fprintf(stream, "xtc_threshold: %f # default: 0.1\n", sparams.xtc_threshold);
2162-
fprintf(stream, "typ_p: %f # default: 1.0\n", sparams.typ_p);
2163-
fprintf(stream, "verbose_prompt: %s # default: false\n", params.verbose_prompt ? "true" : "false");
2164-
fprintf(stream, "display_prompt: %s # default: true\n", params.display_prompt ? "true" : "false");
2165-
}

‎common/sampling.cpp‎

Lines changed: 1 addition & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -167,23 +167,11 @@ struct common_sampler * common_sampler_init(const struct llama_model * model, co
167167
};
168168

169169
llama_sampler_chain_add(result->chain,
170-
llama_sampler_init_logit_bias(
170+
llama_sampler_init_logit_bias(
171171
llama_n_vocab(model),
172172
params.logit_bias.size(),
173173
params.logit_bias.data()));
174174

175-
llama_sampler_chain_add(result->chain,
176-
llama_sampler_init_penalties(
177-
llama_n_vocab (model),
178-
llama_token_eos(model),
179-
llama_token_nl (model),
180-
params.penalty_last_n,
181-
params.penalty_repeat,
182-
params.penalty_freq,
183-
params.penalty_present,
184-
params.penalize_nl,
185-
params.ignore_eos));
186-
187175
if (params.mirostat == 0) {
188176
for (const auto & cnstr : params.samplers) {
189177
switch (cnstr) {

0 commit comments

Comments
 (0)