From d34633d8db6c2e400355de4862cd699154ecc73f Mon Sep 17 00:00:00 2001 From: John <78893154+cmp-nct@users.noreply.github.com> Date: Wed, 10 Jan 2024 14:37:09 +0100 Subject: [PATCH 1/4] clip : support more quantization types (#4846) Uses ggml functions instead of hardcoded names and adds support to quantize into the modern Q-K variants. This is just the bare minimum to get k-types working - a more refined choice of types would be needed to get best quality on low quantizations. I ran a few tests, it doesn't break anything I could notice and a Q6_K ViT works almost as well as Q8_0 but 3 times the inference speed. --- examples/llava/clip.cpp | 62 ++++++++++++++++------------------------- 1 file changed, 24 insertions(+), 38 deletions(-) diff --git a/examples/llava/clip.cpp b/examples/llava/clip.cpp index cfb79e789..2ae8853d3 100644 --- a/examples/llava/clip.cpp +++ b/examples/llava/clip.cpp @@ -126,24 +126,7 @@ static struct ggml_tensor * get_tensor(struct ggml_context * ctx, const std::str } static std::string get_ftype(int ftype) { - switch (ftype) { - case 0: - return "f32"; - case 1: - return "f16"; - case 2: - return "q4_0"; - case 3: - return "q4_1"; - case 6: - return "q5_0"; - case 7: - return "q5_1"; - case 8: - return "q8_0"; - default: - throw std::runtime_error(format("%s: Unrecognized file type: %d\n", __func__, ftype)); - } + return ggml_type_name(static_cast(ftype)); } // @@ -533,6 +516,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) { buffer_size += n_tensors * 128 /* CLIP PADDING */; clip_ctx * new_clip = new clip_ctx; + #ifdef GGML_USE_CUBLAS new_clip->backend = ggml_backend_cuda_init(0); printf("%s: CLIP using CUDA backend\n", __func__); @@ -543,6 +527,7 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) { printf("%s: CLIP using Metal backend\n", __func__); #endif + if (!new_clip->backend) { new_clip->backend = ggml_backend_cpu_init(); printf("%s: CLIP using CPU backend\n", __func__); @@ -931,26 +916,8 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i ggml_type type = GGML_TYPE_Q4_1; - switch (itype) { - case 2: - type = GGML_TYPE_Q4_0; - break; - case 3: - type = GGML_TYPE_Q4_1; - break; - case 6: - type = GGML_TYPE_Q5_0; - break; - case 7: - type = GGML_TYPE_Q5_1; - break; - case 8: - type = GGML_TYPE_Q8_0; - break; - default: - fprintf(stderr, "%s: invalid quantization type %d\n", __func__, itype); - return false; - }; + assert(itype < GGML_TYPE_COUNT); + type = static_cast(itype); auto * ctx_clip = clip_model_load(fname_inp, 2); @@ -1010,6 +977,10 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i if (quantize) { new_type = type; + if (new_type >= GGML_TYPE_Q2_K && name.find("embd") != std::string::npos) { + new_type = GGML_TYPE_Q8_0; // ggml_get_rows needs non K type + // fprintf(stderr, "%s: quantizing %s to %s\n", __func__, name.c_str(), ggml_type_name(new_type)); + } const size_t n_elms = ggml_nelements(cur); float * f32_data; @@ -1054,6 +1025,21 @@ bool clip_model_quantize(const char * fname_inp, const char * fname_out, const i case GGML_TYPE_Q8_0: { new_size = ggml_quantize_q8_0(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); } break; + case GGML_TYPE_Q2_K: { + new_size = ggml_quantize_q2_K(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); + } break; + case GGML_TYPE_Q3_K: { + new_size = ggml_quantize_q3_K(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); + } break; + case GGML_TYPE_Q4_K: { + new_size = ggml_quantize_q4_K(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); + } break; + case GGML_TYPE_Q5_K: { + new_size = ggml_quantize_q5_K(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); + } break; + case GGML_TYPE_Q6_K: { + new_size = ggml_quantize_q6_K(f32_data, new_data, n_elms, cur->ne[0], hist_cur.data()); + } break; default: { fprintf(stderr, "%s: unsupported quantization type %d\n", __func__, new_type); return false; From 329ff615699d32f596d4ebf8baba654c30064e0d Mon Sep 17 00:00:00 2001 From: Austin <77757836+teleprint-me@users.noreply.github.com> Date: Wed, 10 Jan 2024 08:39:09 -0500 Subject: [PATCH 2/4] llama : recognize 1B phi models (#4847) This update categorizes models with 24 layers as MODEL_1B, ensuring compatibility with different Phi model variants without impacting existing Phi-2 model functionality. --- llama.cpp | 1 + 1 file changed, 1 insertion(+) diff --git a/llama.cpp b/llama.cpp index 8e0717db9..0f09d0c2b 100644 --- a/llama.cpp +++ b/llama.cpp @@ -2829,6 +2829,7 @@ static void llm_load_hparams( ml.get_key(LLM_KV_ATTENTION_LAYERNORM_EPS, hparams.f_norm_eps); switch (hparams.n_layer) { + case 24: model.type = e_model::MODEL_1B; break; case 32: model.type = e_model::MODEL_3B; break; default: model.type = e_model::MODEL_UNKNOWN; } From 57d016ba2d46a6e22517a31a75cebb48f9e234b6 Mon Sep 17 00:00:00 2001 From: Brian Date: Thu, 11 Jan 2024 01:09:53 +1100 Subject: [PATCH 3/4] llama : add additional suffixes for model params (#4834) * llm_load_print_meta: Add additional suffixs for model params * Update llama.cpp model param log remove unneeded comments and convert from > to >= --- llama.cpp | 10 +++++++++- 1 file changed, 9 insertions(+), 1 deletion(-) diff --git a/llama.cpp b/llama.cpp index 0f09d0c2b..e1f1932ba 100644 --- a/llama.cpp +++ b/llama.cpp @@ -3146,7 +3146,15 @@ static void llm_load_print_meta(llama_model_loader & ml, llama_model & model) { LLAMA_LOG_INFO("%s: rope_finetuned = %s\n", __func__, hparams.rope_finetuned ? "yes" : "unknown"); LLAMA_LOG_INFO("%s: model type = %s\n", __func__, llama_model_type_name(model.type)); LLAMA_LOG_INFO("%s: model ftype = %s\n", __func__, llama_model_ftype_name(model.ftype).c_str()); - LLAMA_LOG_INFO("%s: model params = %.2f B\n", __func__, ml.n_elements*1e-9); + if (ml.n_elements >= 1e12) { + LLAMA_LOG_INFO("%s: model params = %.2f T\n", __func__, ml.n_elements*1e-12); + } else if (ml.n_elements >= 1e9) { + LLAMA_LOG_INFO("%s: model params = %.2f B\n", __func__, ml.n_elements*1e-9); + } else if (ml.n_elements >= 1e6) { + LLAMA_LOG_INFO("%s: model params = %.2f M\n", __func__, ml.n_elements*1e-6); + } else { + LLAMA_LOG_INFO("%s: model params = %.2f K\n", __func__, ml.n_elements*1e-3); + } if (ml.n_bytes < GiB) { LLAMA_LOG_INFO("%s: model size = %.2f MiB (%.2f BPW) \n", __func__, ml.n_bytes/1024.0/1024.0, ml.n_bytes*8.0/ml.n_elements); } else { From cd108e641dbdedd8c5641c4cec1762f751f38136 Mon Sep 17 00:00:00 2001 From: Behnam M <58621210+ibehnam@users.noreply.github.com> Date: Wed, 10 Jan 2024 14:56:05 -0500 Subject: [PATCH 4/4] server : add a `/health` endpoint (#4860) * added /health endpoint to the server * added comments on the additional /health endpoint * Better handling of server state When the model is being loaded, the server state is `LOADING_MODEL`. If model-loading fails, the server state becomes `ERROR`, otherwise it becomes `READY`. The `/health` endpoint provides more granular messages now according to the server_state value. * initialized server_state * fixed a typo * starting http server before initializing the model * Update server.cpp * Update server.cpp * fixes * fixes * fixes * made ServerState atomic and turned two-line spaces into one-line --- examples/server/server.cpp | 199 +++++++++++++++++++++---------------- 1 file changed, 113 insertions(+), 86 deletions(-) diff --git a/examples/server/server.cpp b/examples/server/server.cpp index 6c7fcd176..1cca634d5 100644 --- a/examples/server/server.cpp +++ b/examples/server/server.cpp @@ -26,6 +26,7 @@ #include #include #include +#include #ifndef SERVER_VERBOSE #define SERVER_VERBOSE 1 @@ -146,6 +147,12 @@ static std::vector base64_decode(const std::string & encoded_string) // parallel // +enum ServerState { + LOADING_MODEL, // Server is starting up, model not fully loaded yet + READY, // Server is ready and model is loaded + ERROR // An error occurred, load_model failed +}; + enum task_type { COMPLETION_TASK, CANCEL_TASK @@ -2453,7 +2460,6 @@ static void server_params_parse(int argc, char **argv, server_params &sparams, } } - static std::string random_string() { static const std::string str("0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz"); @@ -2790,15 +2796,117 @@ int main(int argc, char **argv) {"system_info", llama_print_system_info()}, }); - // load the model - if (!llama.load_model(params)) + httplib::Server svr; + + std::atomic server_state{LOADING_MODEL}; + + svr.set_default_headers({{"Server", "llama.cpp"}, + {"Access-Control-Allow-Origin", "*"}, + {"Access-Control-Allow-Headers", "content-type"}}); + + svr.Get("/health", [&](const httplib::Request&, httplib::Response& res) { + ServerState current_state = server_state.load(); + switch(current_state) { + case READY: + res.set_content(R"({"status": "ok"})", "application/json"); + res.status = 200; // HTTP OK + break; + case LOADING_MODEL: + res.set_content(R"({"status": "loading model"})", "application/json"); + res.status = 503; // HTTP Service Unavailable + break; + case ERROR: + res.set_content(R"({"status": "error", "error": "Model failed to load"})", "application/json"); + res.status = 500; // HTTP Internal Server Error + break; + } + }); + + svr.set_logger(log_server_request); + + svr.set_exception_handler([](const httplib::Request &, httplib::Response &res, std::exception_ptr ep) + { + const char fmt[] = "500 Internal Server Error\n%s"; + char buf[BUFSIZ]; + try + { + std::rethrow_exception(std::move(ep)); + } + catch (std::exception &e) + { + snprintf(buf, sizeof(buf), fmt, e.what()); + } + catch (...) + { + snprintf(buf, sizeof(buf), fmt, "Unknown Exception"); + } + res.set_content(buf, "text/plain; charset=utf-8"); + res.status = 500; + }); + + svr.set_error_handler([](const httplib::Request &, httplib::Response &res) + { + if (res.status == 401) + { + res.set_content("Unauthorized", "text/plain; charset=utf-8"); + } + if (res.status == 400) + { + res.set_content("Invalid request", "text/plain; charset=utf-8"); + } + else if (res.status == 404) + { + res.set_content("File Not Found", "text/plain; charset=utf-8"); + res.status = 404; + } + }); + + // set timeouts and change hostname and port + svr.set_read_timeout (sparams.read_timeout); + svr.set_write_timeout(sparams.write_timeout); + + if (!svr.bind_to_port(sparams.hostname, sparams.port)) { + fprintf(stderr, "\ncouldn't bind to server socket: hostname=%s port=%d\n\n", sparams.hostname.c_str(), sparams.port); return 1; } - llama.initialize(); + // Set the base directory for serving static files + svr.set_base_dir(sparams.public_path); - httplib::Server svr; + // to make it ctrl+clickable: + LOG_TEE("\nllama server listening at http://%s:%d\n\n", sparams.hostname.c_str(), sparams.port); + + std::unordered_map log_data; + log_data["hostname"] = sparams.hostname; + log_data["port"] = std::to_string(sparams.port); + + if (!sparams.api_key.empty()) { + log_data["api_key"] = "api_key: ****" + sparams.api_key.substr(sparams.api_key.length() - 4); + } + + LOG_INFO("HTTP server listening", log_data); + // run the HTTP server in a thread - see comment below + std::thread t([&]() + { + if (!svr.listen_after_bind()) + { + server_state.store(ERROR); + return 1; + } + + return 0; + }); + + // load the model + if (!llama.load_model(params)) + { + server_state.store(ERROR); + return 1; + } else { + llama.initialize(); + server_state.store(READY); + } // Middleware for API key validation auto validate_api_key = [&sparams](const httplib::Request &req, httplib::Response &res) -> bool { @@ -2826,10 +2934,6 @@ int main(int argc, char **argv) return false; }; - svr.set_default_headers({{"Server", "llama.cpp"}, - {"Access-Control-Allow-Origin", "*"}, - {"Access-Control-Allow-Headers", "content-type"}}); - // this is only called if no index.html is found in the public --path svr.Get("/", [](const httplib::Request &, httplib::Response &res) { @@ -2937,8 +3041,6 @@ int main(int argc, char **argv) } }); - - svr.Get("/v1/models", [¶ms](const httplib::Request&, httplib::Response& res) { std::time_t t = std::time(0); @@ -3157,81 +3259,6 @@ int main(int argc, char **argv) return res.set_content(result.result_json.dump(), "application/json; charset=utf-8"); }); - svr.set_logger(log_server_request); - - svr.set_exception_handler([](const httplib::Request &, httplib::Response &res, std::exception_ptr ep) - { - const char fmt[] = "500 Internal Server Error\n%s"; - char buf[BUFSIZ]; - try - { - std::rethrow_exception(std::move(ep)); - } - catch (std::exception &e) - { - snprintf(buf, sizeof(buf), fmt, e.what()); - } - catch (...) - { - snprintf(buf, sizeof(buf), fmt, "Unknown Exception"); - } - res.set_content(buf, "text/plain; charset=utf-8"); - res.status = 500; - }); - - svr.set_error_handler([](const httplib::Request &, httplib::Response &res) - { - if (res.status == 401) - { - res.set_content("Unauthorized", "text/plain; charset=utf-8"); - } - if (res.status == 400) - { - res.set_content("Invalid request", "text/plain; charset=utf-8"); - } - else if (res.status == 404) - { - res.set_content("File Not Found", "text/plain; charset=utf-8"); - res.status = 404; - } - }); - - // set timeouts and change hostname and port - svr.set_read_timeout (sparams.read_timeout); - svr.set_write_timeout(sparams.write_timeout); - - if (!svr.bind_to_port(sparams.hostname, sparams.port)) - { - fprintf(stderr, "\ncouldn't bind to server socket: hostname=%s port=%d\n\n", sparams.hostname.c_str(), sparams.port); - return 1; - } - - // Set the base directory for serving static files - svr.set_base_dir(sparams.public_path); - - // to make it ctrl+clickable: - LOG_TEE("\nllama server listening at http://%s:%d\n\n", sparams.hostname.c_str(), sparams.port); - - std::unordered_map log_data; - log_data["hostname"] = sparams.hostname; - log_data["port"] = std::to_string(sparams.port); - - if (!sparams.api_key.empty()) { - log_data["api_key"] = "api_key: ****" + sparams.api_key.substr(sparams.api_key.length() - 4); - } - - LOG_INFO("HTTP server listening", log_data); - // run the HTTP server in a thread - see comment below - std::thread t([&]() - { - if (!svr.listen_after_bind()) - { - return 1; - } - - return 0; - }); - // GG: if I put the main loop inside a thread, it crashes on the first request when build in Debug!? // "Bus error: 10" - this is on macOS, it does not crash on Linux //std::thread t2([&]()