npm - @fugood/llama.node - Versions diffs - 0.3.1 → 0.3.3 - Mend

@fugood/llama.node 0.3.1 → 0.3.3

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (252) hide show

package/CMakeLists.txt +1 -8
package/bin/darwin/arm64/llama-node.node +0 -0
package/bin/darwin/x64/llama-node.node +0 -0
package/bin/linux/arm64/llama-node.node +0 -0
package/bin/linux/x64/llama-node.node +0 -0
package/bin/linux-vulkan/arm64/llama-node.node +0 -0
package/bin/linux-vulkan/x64/llama-node.node +0 -0
package/bin/win32/arm64/llama-node.node +0 -0
package/bin/win32/arm64/node.lib +0 -0
package/bin/win32/x64/llama-node.node +0 -0
package/bin/win32/x64/node.lib +0 -0
package/bin/win32-vulkan/arm64/llama-node.node +0 -0
package/bin/win32-vulkan/arm64/node.lib +0 -0
package/bin/win32-vulkan/x64/llama-node.node +0 -0
package/bin/win32-vulkan/x64/node.lib +0 -0
package/package.json +4 -2
package/src/DetokenizeWorker.cpp +1 -1
package/src/EmbeddingWorker.cpp +2 -2
package/src/LlamaCompletionWorker.cpp +10 -10
package/src/LlamaCompletionWorker.h +2 -2
package/src/LlamaContext.cpp +14 -17
package/src/TokenizeWorker.cpp +1 -1
package/src/common.hpp +5 -4
package/src/llama.cpp/.github/workflows/build.yml +137 -29
package/src/llama.cpp/.github/workflows/close-issue.yml +5 -0
package/src/llama.cpp/.github/workflows/docker.yml +46 -34
package/src/llama.cpp/.github/workflows/nix-ci-aarch64.yml +7 -0
package/src/llama.cpp/.github/workflows/nix-ci.yml +7 -0
package/src/llama.cpp/.github/workflows/python-check-requirements.yml +2 -4
package/src/llama.cpp/.github/workflows/python-type-check.yml +3 -1
package/src/llama.cpp/.github/workflows/server.yml +7 -0
package/src/llama.cpp/CMakeLists.txt +26 -11
package/src/llama.cpp/cmake/arm64-apple-clang.cmake +16 -0
package/src/llama.cpp/common/CMakeLists.txt +10 -10
package/src/llama.cpp/common/arg.cpp +2041 -0
package/src/llama.cpp/common/arg.h +77 -0
package/src/llama.cpp/common/common.cpp +523 -1861
package/src/llama.cpp/common/common.h +234 -106
package/src/llama.cpp/common/console.cpp +3 -0
package/src/llama.cpp/common/json-schema-to-grammar.cpp +1 -1
package/src/llama.cpp/common/log.cpp +401 -0
package/src/llama.cpp/common/log.h +66 -698
package/src/llama.cpp/common/ngram-cache.cpp +39 -36
package/src/llama.cpp/common/ngram-cache.h +19 -19
package/src/llama.cpp/common/sampling.cpp +356 -350
package/src/llama.cpp/common/sampling.h +62 -139
package/src/llama.cpp/common/stb_image.h +5990 -6398
package/src/llama.cpp/docs/build.md +72 -17
package/src/llama.cpp/examples/CMakeLists.txt +1 -2
package/src/llama.cpp/examples/batched/batched.cpp +49 -65
package/src/llama.cpp/examples/batched-bench/batched-bench.cpp +42 -53
package/src/llama.cpp/examples/convert-llama2c-to-ggml/convert-llama2c-to-ggml.cpp +55 -52
package/src/llama.cpp/examples/cvector-generator/cvector-generator.cpp +22 -22
package/src/llama.cpp/examples/cvector-generator/pca.hpp +3 -13
package/src/llama.cpp/examples/embedding/embedding.cpp +147 -91
package/src/llama.cpp/examples/eval-callback/eval-callback.cpp +37 -37
package/src/llama.cpp/examples/export-lora/export-lora.cpp +39 -38
package/src/llama.cpp/examples/gbnf-validator/gbnf-validator.cpp +14 -39
package/src/llama.cpp/examples/{baby-llama → gen-docs}/CMakeLists.txt +2 -2
package/src/llama.cpp/examples/gen-docs/gen-docs.cpp +83 -0
package/src/llama.cpp/examples/gguf-split/gguf-split.cpp +58 -39
package/src/llama.cpp/examples/gritlm/gritlm.cpp +46 -39
package/src/llama.cpp/examples/imatrix/imatrix.cpp +75 -69
package/src/llama.cpp/examples/infill/infill.cpp +131 -192
package/src/llama.cpp/examples/llama-bench/llama-bench.cpp +276 -178
package/src/llama.cpp/examples/llama.android/llama/build.gradle.kts +1 -0
package/src/llama.cpp/examples/llama.android/llama/src/main/cpp/llama-android.cpp +40 -36
package/src/llama.cpp/examples/llava/CMakeLists.txt +7 -0
package/src/llama.cpp/examples/llava/clip.cpp +686 -150
package/src/llama.cpp/examples/llava/clip.h +11 -2
package/src/llama.cpp/examples/llava/llava-cli.cpp +60 -71
package/src/llama.cpp/examples/llava/llava.cpp +146 -26
package/src/llama.cpp/examples/llava/llava.h +2 -3
package/src/llama.cpp/examples/llava/minicpmv-cli.cpp +323 -0
package/src/llama.cpp/examples/llava/requirements.txt +1 -0
package/src/llama.cpp/examples/lookahead/lookahead.cpp +55 -56
package/src/llama.cpp/examples/lookup/lookup-create.cpp +15 -13
package/src/llama.cpp/examples/lookup/lookup-merge.cpp +4 -4
package/src/llama.cpp/examples/lookup/lookup-stats.cpp +34 -33
package/src/llama.cpp/examples/lookup/lookup.cpp +60 -63
package/src/llama.cpp/examples/main/main.cpp +216 -313
package/src/llama.cpp/examples/parallel/parallel.cpp +58 -59
package/src/llama.cpp/examples/passkey/passkey.cpp +53 -61
package/src/llama.cpp/examples/perplexity/perplexity.cpp +277 -311
package/src/llama.cpp/examples/quantize/CMakeLists.txt +1 -1
package/src/llama.cpp/examples/quantize/quantize.cpp +27 -9
package/src/llama.cpp/examples/quantize-stats/quantize-stats.cpp +12 -12
package/src/llama.cpp/examples/retrieval/retrieval.cpp +57 -52
package/src/llama.cpp/examples/rpc/rpc-server.cpp +27 -2
package/src/llama.cpp/examples/save-load-state/save-load-state.cpp +60 -46
package/src/llama.cpp/examples/server/CMakeLists.txt +7 -18
package/src/llama.cpp/examples/server/server.cpp +1347 -1531
package/src/llama.cpp/examples/server/tests/requirements.txt +2 -1
package/src/llama.cpp/examples/server/utils.hpp +396 -107
package/src/llama.cpp/examples/simple/CMakeLists.txt +1 -1
package/src/llama.cpp/examples/simple/simple.cpp +132 -106
package/src/llama.cpp/examples/simple-chat/CMakeLists.txt +5 -0
package/src/llama.cpp/examples/simple-chat/simple-chat.cpp +197 -0
package/src/llama.cpp/examples/speculative/speculative.cpp +153 -124
package/src/llama.cpp/examples/sycl/run-llama2.sh +10 -19
package/src/llama.cpp/examples/sycl/win-run-llama2.bat +1 -1
package/src/llama.cpp/examples/tokenize/tokenize.cpp +27 -29
package/src/llama.cpp/ggml/CMakeLists.txt +29 -12
package/src/llama.cpp/ggml/include/ggml-alloc.h +3 -3
package/src/llama.cpp/ggml/include/ggml-amx.h +25 -0
package/src/llama.cpp/ggml/include/ggml-backend.h +166 -68
package/src/llama.cpp/ggml/include/ggml-blas.h +5 -3
package/src/llama.cpp/ggml/include/ggml-cann.h +17 -19
package/src/llama.cpp/ggml/include/ggml-cpp.h +38 -0
package/src/llama.cpp/ggml/include/ggml-cpu.h +177 -0
package/src/llama.cpp/ggml/include/ggml-cuda.h +17 -17
package/src/llama.cpp/ggml/include/ggml-kompute.h +7 -3
package/src/llama.cpp/ggml/include/ggml-metal.h +13 -12
package/src/llama.cpp/ggml/include/ggml-opt.h +216 -0
package/src/llama.cpp/ggml/include/ggml-rpc.h +9 -5
package/src/llama.cpp/ggml/include/ggml-sycl.h +18 -11
package/src/llama.cpp/ggml/include/ggml-vulkan.h +10 -8
package/src/llama.cpp/ggml/include/ggml.h +272 -505
package/src/llama.cpp/ggml/src/CMakeLists.txt +69 -1110
package/src/llama.cpp/ggml/src/ggml-aarch64.c +52 -2116
package/src/llama.cpp/ggml/src/ggml-aarch64.h +0 -20
package/src/llama.cpp/ggml/src/ggml-alloc.c +29 -27
package/src/llama.cpp/ggml/src/ggml-amx/CMakeLists.txt +107 -0
package/src/llama.cpp/ggml/src/ggml-amx/common.h +94 -0
package/src/llama.cpp/ggml/src/ggml-amx/ggml-amx.cpp +446 -0
package/src/llama.cpp/ggml/src/ggml-amx/mmq.cpp +2510 -0
package/src/llama.cpp/ggml/src/ggml-amx/mmq.h +17 -0
package/src/llama.cpp/ggml/src/ggml-backend-impl.h +144 -81
package/src/llama.cpp/ggml/src/ggml-backend-reg.cpp +195 -0
package/src/llama.cpp/ggml/src/{ggml-backend.c → ggml-backend.cpp} +394 -635
package/src/llama.cpp/ggml/src/ggml-blas/CMakeLists.txt +91 -0
package/src/llama.cpp/ggml/src/{ggml-blas.cpp → ggml-blas/ggml-blas.cpp} +217 -70
package/src/llama.cpp/ggml/src/ggml-cann/CMakeLists.txt +46 -0
package/src/llama.cpp/ggml/src/ggml-cann/acl_tensor.cpp +4 -27
package/src/llama.cpp/ggml/src/ggml-cann/acl_tensor.h +32 -4
package/src/llama.cpp/ggml/src/ggml-cann/aclnn_ops.cpp +179 -41
package/src/llama.cpp/ggml/src/ggml-cann/common.h +1 -0
package/src/llama.cpp/ggml/src/{ggml-cann.cpp → ggml-cann/ggml-cann.cpp} +458 -353
package/src/llama.cpp/ggml/src/ggml-cann/kernels/CMakeLists.txt +2 -1
package/src/llama.cpp/ggml/src/ggml-cann/kernels/ascendc_kernels.h +2 -0
package/src/llama.cpp/ggml/src/ggml-cann/kernels/quantize_float_to_q4_0.cpp +278 -0
package/src/llama.cpp/ggml/src/ggml-common.h +20 -0
package/src/llama.cpp/ggml/src/ggml-cpu/CMakeLists.txt +261 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-aarch64.c +3560 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-aarch64.h +30 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-impl.h +371 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-quants.c +10822 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu-quants.h +63 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.c +13970 -0
package/src/llama.cpp/ggml/src/ggml-cpu/ggml-cpu.cpp +663 -0
package/src/llama.cpp/ggml/src/ggml-cpu/llamafile/sgemm.cpp +1885 -0
package/src/llama.cpp/ggml/src/ggml-cuda/CMakeLists.txt +155 -0
package/src/llama.cpp/ggml/src/ggml-cuda/vendors/cuda.h +14 -0
package/src/llama.cpp/ggml/src/ggml-cuda/vendors/hip.h +178 -0
package/src/llama.cpp/ggml/src/ggml-cuda/vendors/musa.h +134 -0
package/src/llama.cpp/ggml/src/ggml-hip/CMakeLists.txt +106 -0
package/src/llama.cpp/ggml/src/ggml-impl.h +380 -584
package/src/llama.cpp/ggml/src/ggml-kompute/CMakeLists.txt +162 -0
package/src/llama.cpp/ggml/src/{ggml-kompute.cpp → ggml-kompute/ggml-kompute.cpp} +233 -87
package/src/llama.cpp/ggml/src/ggml-metal/CMakeLists.txt +108 -0
package/src/llama.cpp/ggml/src/ggml-metal/ggml-metal-impl.h +249 -0
package/src/llama.cpp/ggml/src/ggml-musa/CMakeLists.txt +100 -0
package/src/llama.cpp/ggml/src/ggml-opt.cpp +867 -0
package/src/llama.cpp/ggml/src/ggml-quants.c +369 -9994
package/src/llama.cpp/ggml/src/ggml-quants.h +78 -110
package/src/llama.cpp/ggml/src/ggml-rpc/CMakeLists.txt +11 -0
package/src/llama.cpp/ggml/src/{ggml-rpc.cpp → ggml-rpc/ggml-rpc.cpp} +560 -335
package/src/llama.cpp/ggml/src/ggml-sycl/CMakeLists.txt +81 -0
package/src/llama.cpp/ggml/src/ggml-sycl/backend.hpp +6 -0
package/src/llama.cpp/ggml/src/ggml-sycl/common.cpp +51 -0
package/src/llama.cpp/ggml/src/ggml-sycl/common.hpp +310 -0
package/src/llama.cpp/ggml/src/ggml-sycl/concat.cpp +1 -0
package/src/llama.cpp/ggml/src/ggml-sycl/conv.cpp +99 -0
package/src/llama.cpp/ggml/src/ggml-sycl/conv.hpp +21 -0
package/src/llama.cpp/ggml/src/ggml-sycl/convert.cpp +57 -57
package/src/llama.cpp/ggml/src/ggml-sycl/convert.hpp +1 -1
package/src/llama.cpp/ggml/src/ggml-sycl/dequantize.hpp +106 -106
package/src/llama.cpp/ggml/src/ggml-sycl/dmmv.cpp +4 -4
package/src/llama.cpp/ggml/src/ggml-sycl/dpct/helper.hpp +18 -25
package/src/llama.cpp/ggml/src/ggml-sycl/element_wise.cpp +1011 -0
package/src/llama.cpp/ggml/src/ggml-sycl/element_wise.hpp +76 -0
package/src/llama.cpp/ggml/src/ggml-sycl/gemm.hpp +101 -0
package/src/llama.cpp/ggml/src/{ggml-sycl.cpp → ggml-sycl/ggml-sycl.cpp} +3350 -3980
package/src/llama.cpp/ggml/src/ggml-sycl/im2col.cpp +125 -0
package/src/llama.cpp/ggml/src/ggml-sycl/im2col.hpp +23 -0
package/src/llama.cpp/ggml/src/ggml-sycl/mmvq.cpp +70 -68
package/src/llama.cpp/ggml/src/ggml-sycl/norm.cpp +9 -6
package/src/llama.cpp/ggml/src/ggml-sycl/outprod.cpp +56 -0
package/src/llama.cpp/ggml/src/ggml-sycl/outprod.hpp +11 -0
package/src/llama.cpp/ggml/src/ggml-sycl/presets.hpp +8 -0
package/src/llama.cpp/ggml/src/ggml-sycl/rope.cpp +1 -1
package/src/llama.cpp/ggml/src/ggml-sycl/tsembd.cpp +71 -0
package/src/llama.cpp/ggml/src/ggml-sycl/tsembd.hpp +21 -0
package/src/llama.cpp/ggml/src/ggml-sycl/vecdotq.hpp +4 -4
package/src/llama.cpp/ggml/src/ggml-sycl/wkv6.cpp +138 -0
package/src/llama.cpp/ggml/src/ggml-sycl/wkv6.hpp +10 -0
package/src/llama.cpp/ggml/src/ggml-threading.cpp +12 -0
package/src/llama.cpp/ggml/src/ggml-threading.h +12 -0
package/src/llama.cpp/ggml/src/ggml-vulkan/CMakeLists.txt +78 -0
package/src/llama.cpp/ggml/src/{ggml-vulkan.cpp → ggml-vulkan/ggml-vulkan.cpp} +2034 -1718
package/src/llama.cpp/ggml/src/{vulkan-shaders → ggml-vulkan/vulkan-shaders}/CMakeLists.txt +2 -0
package/src/llama.cpp/ggml/src/{vulkan-shaders → ggml-vulkan/vulkan-shaders}/vulkan-shaders-gen.cpp +152 -185
package/src/llama.cpp/ggml/src/ggml.c +2075 -16579
package/src/llama.cpp/include/llama.h +296 -285
package/src/llama.cpp/models/ggml-vocab-chameleon.gguf.inp +112 -0
package/src/llama.cpp/models/ggml-vocab-chameleon.gguf.out +46 -0
package/src/llama.cpp/pocs/vdot/q8dot.cpp +4 -3
package/src/llama.cpp/pocs/vdot/vdot.cpp +8 -7
package/src/llama.cpp/requirements/requirements-convert_legacy_llama.txt +1 -1
package/src/llama.cpp/src/CMakeLists.txt +2 -1
package/src/llama.cpp/src/llama-grammar.cpp +721 -122
package/src/llama.cpp/src/llama-grammar.h +120 -15
package/src/llama.cpp/src/llama-impl.h +156 -1
package/src/llama.cpp/src/llama-sampling.cpp +2058 -346
package/src/llama.cpp/src/llama-sampling.h +39 -47
package/src/llama.cpp/src/llama-vocab.cpp +390 -127
package/src/llama.cpp/src/llama-vocab.h +60 -20
package/src/llama.cpp/src/llama.cpp +6215 -3263
package/src/llama.cpp/src/unicode-data.cpp +6 -4
package/src/llama.cpp/src/unicode-data.h +4 -4
package/src/llama.cpp/src/unicode.cpp +15 -7
package/src/llama.cpp/tests/CMakeLists.txt +4 -2
package/src/llama.cpp/tests/test-arg-parser.cpp +131 -0
package/src/llama.cpp/tests/test-backend-ops.cpp +1725 -297
package/src/llama.cpp/tests/test-barrier.cpp +94 -0
package/src/llama.cpp/tests/test-chat-template.cpp +9 -5
package/src/llama.cpp/tests/test-grammar-integration.cpp +23 -38
package/src/llama.cpp/tests/test-grammar-parser.cpp +6 -4
package/src/llama.cpp/tests/test-json-schema-to-grammar.cpp +23 -8
package/src/llama.cpp/tests/test-llama-grammar.cpp +9 -8
package/src/llama.cpp/tests/test-log.cpp +39 -0
package/src/llama.cpp/tests/test-opt.cpp +853 -142
package/src/llama.cpp/tests/test-quantize-fns.cpp +28 -19
package/src/llama.cpp/tests/test-quantize-perf.cpp +16 -14
package/src/llama.cpp/tests/test-rope.cpp +2 -1
package/src/llama.cpp/tests/test-sampling.cpp +226 -142
package/src/llama.cpp/tests/test-tokenizer-0.cpp +56 -36
package/src/llama.cpp/tests/test-tokenizer-1-bpe.cpp +5 -5
package/src/llama.cpp/tests/test-tokenizer-1-spm.cpp +5 -5
package/patches/llama.patch +0 -22
package/src/llama.cpp/.github/workflows/bench.yml +0 -310
package/src/llama.cpp/common/grammar-parser.cpp +0 -536
package/src/llama.cpp/common/grammar-parser.h +0 -29
package/src/llama.cpp/common/train.cpp +0 -1513
package/src/llama.cpp/common/train.h +0 -233
package/src/llama.cpp/examples/baby-llama/baby-llama.cpp +0 -1640
package/src/llama.cpp/examples/benchmark/CMakeLists.txt +0 -6
package/src/llama.cpp/examples/benchmark/benchmark-matmult.cpp +0 -275
package/src/llama.cpp/ggml/src/llamafile/sgemm.cpp +0 -1027
package/src/llama.cpp/tests/test-grad0.cpp +0 -1566
/package/src/llama.cpp/ggml/{cmake → src/ggml-cpu/cmake}/FindSIMD.cmake +0 -0
/package/src/llama.cpp/ggml/src/{llamafile → ggml-cpu/llamafile}/sgemm.h +0 -0

package/src/llama.cpp/examples/lookup/lookup.cpp CHANGED Viewed

@@ -1,72 +1,68 @@
+#include "arg.h"
 #include "ggml.h"
-#include "llama.h"
 #include "common.h"
 #include "ngram-cache.h"
+#include "sampling.h"
+#include "log.h"
+#include "llama.h"
-#include <cmath>
 #include <cstdint>
 #include <cstdio>
 #include <fstream>
 #include <string>
 #include <vector>
-#include <unordered_map>
 int main(int argc, char ** argv){
-    gpt_params params;
+    common_params params;
-    if (!gpt_params_parse(argc, argv, params)) {
-        gpt_params_print_usage(argc, argv, params);
+    if (!common_params_parse(argc, argv, params, LLAMA_EXAMPLE_LOOKUP)) {
         return 1;
     }
+    common_init();
     // max. number of additional tokens to draft if match is found
     const int n_draft = params.n_draft;
     const bool dump_kv_cache = params.dump_kv_cache;
-#ifndef LOG_DISABLE_LOGS
-    log_set_target(log_filename_generator("lookup", "log"));
-    LOG_TEE("Log start\n");
-    log_dump_cmdline(argc, argv);
-#endif // LOG_DISABLE_LOGS
     // init llama.cpp
     llama_backend_init();
     llama_numa_init(params.numa);
-    llama_model * model = NULL;
-    llama_context * ctx = NULL;
     // load the model
-    std::tie(model, ctx) = llama_init_from_gpt_params(params);
+    common_init_result llama_init = common_init_from_params(params);
+    llama_model * model = llama_init.model;
+    llama_context * ctx = llama_init.context;
     // tokenize the prompt
     std::vector<llama_token> inp;
-    inp = ::llama_tokenize(ctx, params.prompt, true, true);
+    inp = common_tokenize(ctx, params.prompt, true, true);
-    llama_ngram_cache ngram_cache_context;
-    llama_ngram_cache ngram_cache_dynamic;
-    llama_ngram_cache ngram_cache_static;
+    common_ngram_cache ngram_cache_context;
+    common_ngram_cache ngram_cache_dynamic;
+    common_ngram_cache ngram_cache_static;
     int64_t t_draft_flat_us = 0;
     int64_t t_draft_us = 0;
     {
         // Fill up context ngram cache with tokens from user input:
         const int64_t t_start_draft_us = ggml_time_us();
-        llama_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, inp.size(), false);
+        common_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, inp.size(), false);
         if (!params.lookup_cache_static.empty()) {
             try {
-                ngram_cache_static = llama_ngram_cache_load(params.lookup_cache_static);
+                ngram_cache_static = common_ngram_cache_load(params.lookup_cache_static);
             } catch (std::ifstream::failure const &) {
-                fprintf(stderr, "error: failed to open static lookup cache: %s", params.lookup_cache_static.c_str());
+                LOG_ERR("failed to open static lookup cache: %s", params.lookup_cache_static.c_str());
                 exit(1);
             }
         }
         if (!params.lookup_cache_dynamic.empty()) {
             try {
-                ngram_cache_dynamic = llama_ngram_cache_load(params.lookup_cache_dynamic);
+                ngram_cache_dynamic = common_ngram_cache_load(params.lookup_cache_dynamic);
             } catch (std::ifstream::failure const &) {} // if the file does not exist it will simply be created at the end of the program
         }
@@ -77,14 +73,14 @@ int main(int argc, char ** argv){
     const int max_tokens_list_size = max_context_size - 4;
     if ((int) inp.size() > max_tokens_list_size) {
-        fprintf(stderr, "%s: error: prompt too long (%d tokens, max %d)\n", __func__, (int) inp.size(), max_tokens_list_size);
+        LOG_ERR("%s: prompt too long (%d tokens, max %d)\n", __func__, (int) inp.size(), max_tokens_list_size);
         return 1;
     }
-    fprintf(stderr, "\n\n");
+    LOG("\n\n");
     for (auto id : inp) {
-        fprintf(stderr, "%s", llama_token_to_piece(ctx, id).c_str());
+        LOG("%s", common_token_to_piece(ctx, id).c_str());
     }
     fflush(stderr);
@@ -93,8 +89,8 @@ int main(int argc, char ** argv){
     const auto t_enc_start = ggml_time_us();
-    llama_decode(ctx, llama_batch_get_one( inp.data(), n_input - 1, 0,           0));
-    llama_decode(ctx, llama_batch_get_one(&inp.back(),           1, n_input - 1, 0));
+    llama_decode(ctx, llama_batch_get_one( inp.data(), n_input - 1));
+    llama_decode(ctx, llama_batch_get_one(&inp.back(),           1));
     const auto t_enc_end = ggml_time_us();
@@ -106,7 +102,7 @@ int main(int argc, char ** argv){
     bool has_eos = false;
-    struct llama_sampling_context * ctx_sampling = llama_sampling_init(params.sparams);
+    struct common_sampler * smpl = common_sampler_init(model, params.sparams);
     std::vector<llama_token> draft;
@@ -121,23 +117,23 @@ int main(int argc, char ** argv){
         // debug
         if (dump_kv_cache) {
             llama_kv_cache_view_update(ctx, &kvc_view);
-            llama_kv_cache_dump_view_seqs(kvc_view, 40);
+            common_kv_cache_dump_view_seqs(kvc_view, 40);
         }
         // print current draft sequence
-        LOG("drafted %s\n", LOG_TOKENS_TOSTR_PRETTY(ctx, draft).c_str());
+        LOG_DBG("drafted %s\n", string_from(ctx, draft).c_str());
         int i_dft = 0;
         while (true) {
             // sample from the target model
-            llama_token id = llama_sampling_sample(ctx_sampling, ctx, NULL, i_dft);
+            llama_token id = common_sampler_sample(smpl, ctx, i_dft);
-            llama_sampling_accept(ctx_sampling, ctx, id, true);
+            common_sampler_accept(smpl, id, true);
-            const std::string token_str = llama_token_to_piece(ctx, id);
+            const std::string token_str = common_token_to_piece(ctx, id);
             if (!params.use_color) {
-                printf("%s", token_str.c_str());
+                LOG("%s", token_str.c_str());
             }
             if (llama_token_is_eog(model, id)) {
@@ -148,7 +144,7 @@ int main(int argc, char ** argv){
             // check if the target token matches the draft
             if (i_dft < (int) draft.size() && id == draft[i_dft]) {
-                LOG("the sampled target token matches the %dth drafted token (%d, '%s') - accepted\n", i_dft, id, token_str.c_str());
+                LOG_DBG("the sampled target token matches the %dth drafted token (%d, '%s') - accepted\n", i_dft, id, token_str.c_str());
                 ++n_accept;
                 ++n_past;
                 ++i_dft;
@@ -156,25 +152,25 @@ int main(int argc, char ** argv){
                 {
                     // Update context ngram cache with the newly accepted token:
                     const int64_t t_start_draft_us = ggml_time_us();
-                    llama_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, 1, false);
+                    common_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, 1, false);
                     t_draft_us += ggml_time_us() - t_start_draft_us;
                 }
                 if (params.use_color) {
                     // color accepted draft token
-                    printf("\033[34m%s\033[0m", token_str.c_str());
+                    LOG("\033[34m%s\033[0m", token_str.c_str());
                     fflush(stdout);
                 }
                 continue;
             }
             if (params.use_color) {
-                printf("%s", token_str.c_str());
+                LOG("%s", token_str.c_str());
             }
             fflush(stdout);
-            LOG("the sampled target token (%d, '%s') did not match, or we ran out of drafted tokens\n", id, token_str.c_str());
+            LOG_DBG("the sampled target token (%d, '%s') did not match, or we ran out of drafted tokens\n", id, token_str.c_str());
             draft.clear();
             draft.push_back(id);
@@ -182,7 +178,7 @@ int main(int argc, char ** argv){
             {
                 // Update context ngram cache with the newly accepted token:
                 const int64_t t_start_draft_us = ggml_time_us();
-                llama_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, 1, false);
+                common_ngram_cache_update(ngram_cache_context, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, inp, 1, false);
                 t_draft_us += ggml_time_us() - t_start_draft_us;
             }
             break;
@@ -196,18 +192,18 @@ int main(int argc, char ** argv){
         // clean the cache of draft tokens that weren't accepted
         llama_kv_cache_seq_rm(ctx, 0, n_past, -1);
-        llama_batch_clear(batch_tgt);
-        llama_batch_add(batch_tgt, draft[0], n_past, { 0 }, true);
+        common_batch_clear(batch_tgt);
+        common_batch_add(batch_tgt, draft[0], n_past, { 0 }, true);
         // Draft already contains a single token sampled from the model:
         GGML_ASSERT(draft.size() == 1);
         GGML_ASSERT(draft[0] == inp.back());
         const int64_t t_start_draft_us = ggml_time_us();
-        llama_ngram_cache_draft(inp, draft, n_draft, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, ngram_cache_context, ngram_cache_dynamic, ngram_cache_static);
+        common_ngram_cache_draft(inp, draft, n_draft, LLAMA_NGRAM_MIN, LLAMA_NGRAM_MAX, ngram_cache_context, ngram_cache_dynamic, ngram_cache_static);
         for (size_t i = 1; i < draft.size(); ++i) {
-            llama_batch_add(batch_tgt, draft[i], n_past + i, { 0 }, true);
+            common_batch_add(batch_tgt, draft[i], n_past + i, { 0 }, true);
         }
         t_draft_us += ggml_time_us() - t_start_draft_us;
@@ -222,28 +218,29 @@ int main(int argc, char ** argv){
     auto t_dec_end = ggml_time_us();
     // Update dynamic ngram cache with context ngram cache and save it to disk:
-    llama_ngram_cache_merge(ngram_cache_dynamic, ngram_cache_context);
-    llama_ngram_cache_save(ngram_cache_dynamic, params.lookup_cache_dynamic);
+    common_ngram_cache_merge(ngram_cache_dynamic, ngram_cache_context);
+    common_ngram_cache_save(ngram_cache_dynamic, params.lookup_cache_dynamic);
-    LOG_TEE("\n\n");
+    LOG("\n\n");
-    LOG_TEE("encoded %4d tokens in %8.3f seconds, speed: %8.3f t/s\n", n_input,   (t_enc_end - t_enc_start) / 1e6f, inp.size() / ((t_enc_end - t_enc_start) / 1e6f));
-    LOG_TEE("decoded %4d tokens in %8.3f seconds, speed: %8.3f t/s\n", n_predict, (t_dec_end - t_dec_start) / 1e6f, n_predict  / ((t_dec_end - t_dec_start) / 1e6f));
+    LOG_INF("encoded %4d tokens in %8.3f seconds, speed: %8.3f t/s\n", n_input,   (t_enc_end - t_enc_start) / 1e6f, inp.size() / ((t_enc_end - t_enc_start) / 1e6f));
+    LOG_INF("decoded %4d tokens in %8.3f seconds, speed: %8.3f t/s\n", n_predict, (t_dec_end - t_dec_start) / 1e6f, n_predict  / ((t_dec_end - t_dec_start) / 1e6f));
-    LOG_TEE("\n");
-    LOG_TEE("n_draft      = %d\n", n_draft);
-    LOG_TEE("n_predict    = %d\n", n_predict);
-    LOG_TEE("n_drafted    = %d\n", n_drafted);
-    LOG_TEE("t_draft_flat = %.2f ms\n", t_draft_flat_us*1e-3);
-    LOG_TEE("t_draft      = %.2f ms, %.2f us per token, %.2f tokens per second\n",
+    LOG_INF("\n");
+    LOG_INF("n_draft      = %d\n", n_draft);
+    LOG_INF("n_predict    = %d\n", n_predict);
+    LOG_INF("n_drafted    = %d\n", n_drafted);
+    LOG_INF("t_draft_flat = %.2f ms\n", t_draft_flat_us*1e-3);
+    LOG_INF("t_draft      = %.2f ms, %.2f us per token, %.2f tokens per second\n",
             t_draft_us*1e-3, 1.0f*t_draft_us/n_drafted, n_drafted/(1e-6*t_draft_us));
-    LOG_TEE("n_accept     = %d\n", n_accept);
-    LOG_TEE("accept       = %.3f%%\n", 100.0f * n_accept / n_drafted);
+    LOG_INF("n_accept     = %d\n", n_accept);
+    LOG_INF("accept       = %.3f%%\n", 100.0f * n_accept / n_drafted);
+    LOG_INF("\ntarget:\n\n");
+    common_perf_print(ctx, smpl);
-    LOG_TEE("\ntarget:\n");
-    llama_print_timings(ctx);
+    common_sampler_free(smpl);
-    llama_sampling_free(ctx_sampling);
     llama_batch_free(batch_tgt);
     llama_free(ctx);
@@ -251,7 +248,7 @@ int main(int argc, char ** argv){
     llama_backend_free();
-    fprintf(stderr, "\n\n");
+    LOG("\n\n");
     return 0;
 }