Fix bin dir for win ci

specify build type for ctest on windows (#371 )
Add notice about pending change
2026-02-26 14:23:22 +02:00 · 2023-03-22 00:01:08 +02:00 · 2023-03-21 23:34:25 +02:00 · 2023-03-21 22:57:35 +02:00 · 2023-03-21 22:52:27 +02:00 · 2023-03-21 19:47:27 +02:00
23 changed files with 1575 additions and 398 deletions
--- a/.devops/full.Dockerfile
+++ b/.devops/full.Dockerfile
@@ -6,7 +6,7 @@ RUN apt-get update && \
    apt-get install -y build-essential python3 python3-pip

 RUN pip install --upgrade pip setuptools wheel \
-    && pip install torch torchvision torchaudio sentencepiece numpy
+    && pip install numpy requests sentencepiece torch tqdm

 WORKDIR /app

@@ -14,4 +14,4 @@ COPY . .

 RUN make

-ENTRYPOINT ["/app/.devops/tools.sh"]
+ENTRYPOINT ["/app/.devops/tools.sh"]
--- a/.github/ISSUE_TEMPLATE/custom.md
+++ b/.github/ISSUE_TEMPLATE/custom.md
@@ -0,0 +1,198 @@
+---
+name: Custom issue template
+about: Used to report user-related issues with the software
+title: "[User] I encountered a problem .."
+labels: ''
+assignees: ''
+
+---
+
+# Prerequisites
+
+Please answer the following questions for yourself before submitting an issue.
+
+- [ ] I am running the latest code. Development is very rapid so there are no tagged versions as of now.
+- [ ] I carefully followed the [README.md](https://github.com/ggerganov/llama.cpp/blob/master/README.md).
+- [ ] I [searched using keywords relevant to my issue](https://docs.github.com/en/issues/tracking-your-work-with-issues/filtering-and-searching-issues-and-pull-requests) to make sure that I am creating a new issue that is not already open (or closed).
+- [ ] I reviewed the [Discussions](https://github.com/ggerganov/llama.cpp/discussions), and have a new bug or useful enhancement to share.
+
+# Expected Behavior
+
+Please provide a detailed written description of what you were trying to do, and what you expected `lamma.cpp` to do.
+
+# Current Behavior
+
+Please provide a detailed written description of what `lamma.cpp` did, instead. 
+
+# Environment and Context 
+
+Please provide detailed information about your computer setup. This is important in case the issue is not reproducible except for under certain specific conditions.
+
+* Physical (or virtual) hardware you are using, e.g. for Linux:
+
+`$ lscpu`
+
+* Operating System, e.g. for Linux:
+
+`$ uname -a`
+
+* SDK version, e.g. for Linux:
+
+```
+$ python3 --version
+$ make --version
+$ g++ --version
+```
+
+# Models
+
+* The LLaMA models are officially distributed by Facebook and will never be provided through this repository. See this [pull request in Facebook's LLaMA repository](https://github.com/facebookresearch/llama/pull/73/files) if you need to obtain access to the model data.
+* If your issue is with model conversion please verify the `sha256sum` of each of your `consolidated*.pth` and `ggml-model-XXX.bin` files to confirm that you have the correct model data files before logging an issue. [Latest sha256 sums for your reference](https://github.com/ggerganov/llama.cpp/issues/238).
+* If your issue is with model generation quality then please at least scan the following links and papers to understand the limitations of LLaMA models. This is especially important when choosing an appropriate model size and appreciating both the significant and subtle differences between LLaMA models and ChatGPT:
+  * LLaMA:
+    * [Introducing LLaMA: A foundational, 65-billion-parameter large language model](https://ai.facebook.com/blog/large-language-model-llama-meta-ai/)
+    * [LLaMA: Open and Efficient Foundation Language Models](https://arxiv.org/abs/2302.13971)
+  * GPT-3
+    * [Language Models are Few-Shot Learners](https://arxiv.org/abs/2005.14165)
+  * GPT-3.5 / InstructGPT / ChatGPT:
+    * [Aligning language models to follow instructions](https://openai.com/research/instruction-following)
+    * [Training language models to follow instructions with human feedback](https://arxiv.org/abs/2203.02155)
+
+# Failure Information (for bugs)
+
+Please help provide information about the failure if this is a bug. If it is not a bug, please remove the rest of this template.
+
+# Steps to Reproduce
+
+Please provide detailed steps for reproducing the issue. We are not sitting in front of your screen, so the more detail the better.
+
+1. step 1
+2. step 2
+3. step 3
+4. etc.
+
+# Failure Logs
+
+Please include any relevant log snippets or files. If it works under one configuration but not under another, please provide logs for both configurations and their corresponding outputs so it is easy to see where behavior changes.
+
+Also, please try to **avoid using screenshots** if at all possible. Instead, copy/paste the console output and use [Github's markdown](https://docs.github.com/en/get-started/writing-on-github/getting-started-with-writing-and-formatting-on-github/basic-writing-and-formatting-syntax) to cleanly format your logs for easy readability. e.g.
+
+```
+llama.cpp$ git log | head -1
+commit 2af23d30434a677c6416812eea52ccc0af65119c
+
+llama.cpp$ lscpu | egrep "AMD|Flags"
+Vendor ID:                       AuthenticAMD
+Model name:                      AMD Ryzen Threadripper 1950X 16-Core Processor
+Flags:                           fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc rep_good nopl nonstop_tsc cpuid extd_apicid amd_dcm aperfmperf rapl pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_llc mwaitx cpb hw_pstate ssbd ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 xsaves clzero irperf xsaveerptr arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov succor smca sme sev
+Virtualization:                  AMD-V
+
+llama.cpp$ python3 --version
+Python 3.10.9
+
+llama.cpp$ pip list | egrep "torch|numpy|sentencepiece"
+numpy                         1.24.2
+numpydoc                      1.5.0
+sentencepiece                 0.1.97
+torch                         1.13.1
+torchvision                   0.14.1
+
+llama.cpp$ make --version | head -1
+GNU Make 4.3
+
+$ md5sum ./models/65B/ggml-model-q4_0.bin
+dbdd682cce80e2d6e93cefc7449df487  ./models/65B/ggml-model-q4_0.bin
+```
+Here's a run with the Linux command [perf](https://www.brendangregg.com/perf.html)
+
+```
+llama.cpp$ perf stat ./main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p "Please close your issue when it has been answered."
+main: seed = 1679149377
+llama_model_load: loading model from './models/65B/ggml-model-q4_0.bin' - please wait ...
+llama_model_load: n_vocab = 32000
+llama_model_load: n_ctx   = 512
+llama_model_load: n_embd  = 8192
+llama_model_load: n_mult  = 256
+llama_model_load: n_head  = 64
+llama_model_load: n_layer = 80
+llama_model_load: n_rot   = 128
+llama_model_load: f16     = 2
+llama_model_load: n_ff    = 22016
+llama_model_load: n_parts = 8
+llama_model_load: ggml ctx size = 41477.73 MB
+llama_model_load: memory_size =  2560.00 MB, n_mem = 40960
+llama_model_load: loading model part 1/8 from './models/65B/ggml-model-q4_0.bin'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 2/8 from './models/65B/ggml-model-q4_0.bin.1'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 3/8 from './models/65B/ggml-model-q4_0.bin.2'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 4/8 from './models/65B/ggml-model-q4_0.bin.3'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 5/8 from './models/65B/ggml-model-q4_0.bin.4'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 6/8 from './models/65B/ggml-model-q4_0.bin.5'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 7/8 from './models/65B/ggml-model-q4_0.bin.6'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+llama_model_load: loading model part 8/8 from './models/65B/ggml-model-q4_0.bin.7'
+llama_model_load: .......................................................................................... done
+llama_model_load: model size =  4869.09 MB / num tensors = 723
+
+system_info: n_threads = 16 / 32 | AVX = 1 | AVX2 = 1 | AVX512 = 0 | FMA = 1 | NEON = 0 | ARM_FMA = 0 | F16C = 1 | FP16_VA = 0 | WASM_SIMD = 0 | BLAS = 0 | SSE3 = 1 | VSX = 0 | 
+
+main: prompt: 'Please close your issue when it has been answered.'
+main: number of tokens in prompt = 11
+     1 -> ''
+ 12148 -> 'Please'
+  3802 -> ' close'
+   596 -> ' your'
+  2228 -> ' issue'
+   746 -> ' when'
+   372 -> ' it'
+   756 -> ' has'
+  1063 -> ' been'
+  7699 -> ' answered'
+ 29889 -> '.'
+
+sampling parameters: temp = 0.800000, top_k = 40, top_p = 0.950000, repeat_last_n = 64, repeat_penalty = 1.300000
+
+
+Please close your issue when it has been answered.
+@duncan-donut: I'm trying to figure out what kind of "support" you need for this script and why, exactly? Is there a question about how the code works that hasn't already been addressed in one or more comments below this ticket, or are we talking something else entirely like some sorta bugfixing job because your server setup is different from mine??
+I can understand if your site needs to be running smoothly and you need help with a fix of sorts but there should really be nothing wrong here that the code itself could not handle. And given that I'm getting reports about how it works perfectly well on some other servers, what exactly are we talking? A detailed report will do wonders in helping us get this resolved for ya quickly so please take your time and describe the issue(s) you see as clearly & concisely as possible!!
+@duncan-donut: I'm not sure if you have access to cPanel but you could try these instructions. It is worth a shot! Let me know how it goes (or what error message, exactly!) when/if ya give that code a go? [end of text]
+
+
+main: mem per token = 71159620 bytes
+main:     load time = 19309.95 ms
+main:   sample time =   168.62 ms
+main:  predict time = 223895.61 ms / 888.47 ms per token
+main:    total time = 246406.42 ms
+
+ Performance counter stats for './main -m ./models/65B/ggml-model-q4_0.bin -t 16 -n 1024 -p Please close your issue when it has been answered.':
+
+        3636882.89 msec task-clock                #   14.677 CPUs utilized          
+             13509      context-switches          #    3.714 /sec                   
+              2436      cpu-migrations            #    0.670 /sec                   
+          10476679      page-faults               #    2.881 K/sec                  
+    13133115082869      cycles                    #    3.611 GHz                      (16.77%)
+       29314462753      stalled-cycles-frontend   #    0.22% frontend cycles idle     (16.76%)
+    10294402631459      stalled-cycles-backend    #   78.39% backend cycles idle      (16.74%)
+    23479217109614      instructions              #    1.79  insn per cycle         
+                                                  #    0.44  stalled cycles per insn  (16.76%)
+     2353072268027      branches                  #  647.002 M/sec                    (16.77%)
+        1998682780      branch-misses             #    0.08% of all branches          (16.76%)
+
+     247.802177522 seconds time elapsed
+
+    3618.573072000 seconds user
+      18.491698000 seconds sys
+```
--- a/.github/workflows/build.yml
+++ b/.github/workflows/build.yml
@@ -54,6 +54,7 @@ jobs:
          cd build
          cmake ..
          cmake --build . --config Release
+          ctest --output-on-failure

  macOS-latest-make:
    runs-on: macos-latest
@@ -90,6 +91,7 @@ jobs:
          cd build
          cmake ..
          cmake --build . --config Release
+          ctest --output-on-failure

  windows-latest-cmake:
    runs-on: windows-latest
@@ -106,6 +108,7 @@ jobs:
          cd build
          cmake ..
          cmake --build . --config Release
+          ctest -C Release --output-on-failure

      - name: Get commit hash
        id: commit
@@ -116,7 +119,7 @@ jobs:
        id: pack_artifacts
        if: ${{ ( github.event_name == 'push' && github.ref == 'refs/heads/master' ) || github.event.inputs.create_release == 'true' }}
        run: |
-          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-x64.zip .\build\Release\*
+          7z a llama-${{ env.BRANCH_NAME }}-${{ steps.commit.outputs.short }}-bin-win-x64.zip .\build\bin\Release\*

      - name: Create release
        id: create_release
--- a/.github/workflows/docker.yml
+++ b/.github/workflows/docker.yml
@@ -40,7 +40,7 @@ jobs:
        uses: docker/login-action@v2
        with:
          registry: ghcr.io
-          username: ${{ github.actor }}
+          username: ${{ github.repository_owner }}
          password: ${{ secrets.GITHUB_TOKEN }}

      - name: Build and push Docker image (versioned)
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -1,131 +1,252 @@
-cmake_minimum_required(VERSION 3.8)
-project("llama.cpp")
+cmake_minimum_required(VERSION 3.12) # Don't bump this version for no reason
+project("llama.cpp" C CXX)

-set(CMAKE_CXX_STANDARD 20)
-set(CMAKE_CXX_STANDARD_REQUIRED true)
-set(CMAKE_C_STANDARD 11)
-set(THREADS_PREFER_PTHREAD_FLAG ON)
-find_package(Threads REQUIRED)
+set(CMAKE_EXPORT_COMPILE_COMMANDS ON)

 if (NOT XCODE AND NOT MSVC AND NOT CMAKE_BUILD_TYPE)
    set(CMAKE_BUILD_TYPE Release CACHE STRING "Build type" FORCE)
    set_property(CACHE CMAKE_BUILD_TYPE PROPERTY STRINGS "Debug" "Release" "MinSizeRel" "RelWithDebInfo")
 endif()

-option(LLAMA_ALL_WARNINGS            "llama: enable all compiler warnings"                   ON)
-option(LLAMA_ALL_WARNINGS_3RD_PARTY  "llama: enable all compiler warnings in 3rd party libs" OFF)
+set(CMAKE_RUNTIME_OUTPUT_DIRECTORY ${CMAKE_BINARY_DIR}/bin)

-option(LLAMA_SANITIZE_THREAD         "llama: enable thread sanitizer"    OFF)
-option(LLAMA_SANITIZE_ADDRESS        "llama: enable address sanitizer"   OFF)
-option(LLAMA_SANITIZE_UNDEFINED      "llama: enable undefined sanitizer" OFF)
+if(CMAKE_SOURCE_DIR STREQUAL CMAKE_CURRENT_SOURCE_DIR)
+    set(LLAMA_STANDALONE ON)

-if (APPLE)
-    option(LLAMA_NO_ACCELERATE       "llama: disable Accelerate framework" OFF)
-    option(LLAMA_NO_AVX              "llama: disable AVX" OFF)
-    option(LLAMA_NO_AVX2             "llama: disable AVX2" OFF)
-    option(LLAMA_NO_FMA              "llama: disable FMA" OFF)
+    # configure project version
+    # TODO
+else()
+    set(LLAMA_STANDALONE OFF)
 endif()

+if (EMSCRIPTEN)
+    set(BUILD_SHARED_LIBS_DEFAULT OFF)
+
+    option(LLAMA_WASM_SINGLE_FILE "llama: embed WASM inside the generated llama.js" ON)
+else()
+    if (MINGW)
+        set(BUILD_SHARED_LIBS_DEFAULT OFF)
+    else()
+        set(BUILD_SHARED_LIBS_DEFAULT ON)
+    endif()
+endif()
+
+
+#
+# Option list
+#
+
+# general
+option(LLAMA_STATIC                 "llama: static link libraries"                          OFF)
+option(LLAMA_NATIVE                 "llama: enable -march=native flag"                      OFF)
+option(LLAMA_LTO                    "llama: enable link time optimization"                  OFF)
+
+# debug
+option(LLAMA_ALL_WARNINGS           "llama: enable all compiler warnings"                   ON)
+option(LLAMA_ALL_WARNINGS_3RD_PARTY "llama: enable all compiler warnings in 3rd party libs" OFF)
+option(LLAMA_GPROF                  "llama: enable gprof"                                   OFF)
+
+# sanitizers
+option(LLAMA_SANITIZE_THREAD        "llama: enable thread sanitizer"                        OFF)
+option(LLAMA_SANITIZE_ADDRESS       "llama: enable address sanitizer"                       OFF)
+option(LLAMA_SANITIZE_UNDEFINED     "llama: enable undefined sanitizer"                     OFF)
+
+# instruction set specific
+option(LLAMA_AVX                    "llama: enable AVX"                                     ON)
+option(LLAMA_AVX2                   "llama: enable AVX2"                                    ON)
+option(LLAMA_FMA                    "llama: enable FMA"                                     ON)
+
+# 3rd party libs
+option(LLAMA_ACCELERATE             "llama: enable Accelerate framework"                    ON)
+option(LLAMA_OPENBLAS               "llama: use OpenBLAS"                                   OFF)
+
+option(LLAMA_BUILD_TESTS            "llama: build tests"    ${LLAMA_STANDALONE})
+option(LLAMA_BUILD_EXAMPLES         "llama: build examples" ${LLAMA_STANDALONE})
+
+#
+# Compile flags
+#
+
+set(CMAKE_CXX_STANDARD_REQUIRED true)
+set(CMAKE_C_STANDARD_REQUIRED true)
+set(THREADS_PREFER_PTHREAD_FLAG ON)
+find_package(Threads REQUIRED)
+
 if (NOT MSVC)
    if (LLAMA_SANITIZE_THREAD)
-        set(CMAKE_C_FLAGS   "${CMAKE_C_FLAGS}   -fsanitize=thread")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=thread")
+        add_compile_options(-fsanitize=thread)
    endif()

    if (LLAMA_SANITIZE_ADDRESS)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}     -fsanitize=address -fno-omit-frame-pointer")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=address -fno-omit-frame-pointer")
+        add_compile_options(-fsanitize=address -fno-omit-frame-pointer)
    endif()

    if (LLAMA_SANITIZE_UNDEFINED)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS}     -fsanitize=undefined")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} -fsanitize=undefined")
+        add_compile_options(-fsanitize=undefined)
    endif()
 endif()

-if (APPLE AND NOT LLAMA_NO_ACCELERATE)
+if (APPLE AND LLAMA_ACCELERATE)
    find_library(ACCELERATE_FRAMEWORK Accelerate)
    if (ACCELERATE_FRAMEWORK)
        message(STATUS "Accelerate framework found")

-        set(LLAMA_EXTRA_LIBS  ${LLAMA_EXTRA_LIBS}  ${ACCELERATE_FRAMEWORK})
-        set(LLAMA_EXTRA_FLAGS ${LLAMA_EXTRA_FLAGS} -DGGML_USE_ACCELERATE)
+        add_compile_definitions(GGML_USE_ACCELERATE)
+        set(LLAMA_EXTRA_LIBS ${LLAMA_EXTRA_LIBS} ${ACCELERATE_FRAMEWORK})
    else()
        message(WARNING "Accelerate framework not found")
    endif()
 endif()
+if (LLAMA_OPENBLAS)
+    if (LLAMA_STATIC)
+        set(BLA_STATIC ON)
+    endif()
+
+    set(BLA_VENDOR OpenBLAS)
+    find_package(BLAS)
+    if (BLAS_FOUND)
+        message(STATUS "OpenBLAS found")
+
+        add_compile_definitions(GGML_USE_OPENBLAS)
+        add_link_options(${BLAS_LIBRARIES})
+    else()
+        message(WARNING "OpenBLAS not found")
+    endif()
+endif()

 if (LLAMA_ALL_WARNINGS)
    if (NOT MSVC)
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} \
-            -Wall                           \
-            -Wextra                         \
-            -Wpedantic                      \
-            -Wshadow                        \
-            -Wcast-qual                     \
-            -Wstrict-prototypes             \
-            -Wpointer-arith                 \
-            -Wno-unused-function            \
-        ")
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} \
-            -Wall                           \
-            -Wextra                         \
-            -Wpedantic                      \
-            -Wcast-qual                     \
-        ")
+        set(c_flags
+            -Wall
+            -Wextra
+            -Wpedantic
+            -Wshadow
+            -Wcast-qual
+            -Wstrict-prototypes
+            -Wpointer-arith
+            -Wno-unused-function
+        )
+        set(cxx_flags
+            -Wall
+            -Wextra
+            -Wpedantic
+            -Wcast-qual
+        )
    else()
        # todo : msvc
    endif()
+
+    add_compile_options(
+            "$<$<COMPILE_LANGUAGE:C>:${c_flags}>"
+            "$<$<COMPILE_LANGUAGE:CXX>:${cxx_flags}>"
+    )
+
 endif()

-message(STATUS "CMAKE_SYSTEM_PROCESSOR: ${CMAKE_SYSTEM_PROCESSOR}")
-
-if (${CMAKE_SYSTEM_PROCESSOR} MATCHES "arm" OR ${CMAKE_SYSTEM_PROCESSOR} MATCHES "aarch64")
-    message(STATUS "ARM detected")
-else()
-    message(STATUS "x86 detected")
-    if (MSVC)
-        set(CMAKE_CXX_FLAGS "${CMAKE_CXX_FLAGS} /arch:AVX2")
-        set(CMAKE_CXX_FLAGS_RELEASE "${CMAKE_CXX_FLAGS_RELEASE} /arch:AVX2")
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} /arch:AVX2")
+if (LLAMA_LTO)
+    include(CheckIPOSupported)
+    check_ipo_supported(RESULT result OUTPUT output)
+    if (result)
+        set(CMAKE_INTERPROCEDURAL_OPTIMIZATION TRUE)
    else()
-        if(NOT LLAMA_NO_AVX)
-            set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mavx")
-        endif()
-        if(NOT LLAMA_NO_AVX2)
-            set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mavx2")
-        endif()
-        if(NOT LLAMA_NO_FMA)
-            set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mfma")
-        endif()
-        set(CMAKE_C_FLAGS "${CMAKE_C_FLAGS} -mf16c")
+        message(WARNING "IPO is not supported: ${output}")
    endif()
 endif()

-# if (LLAMA_PERF)
-#     set(LLAMA_EXTRA_FLAGS ${LLAMA_EXTRA_FLAGS} -DGGML_PERF)
-# endif()
+# Architecture specific
+# TODO: probably these flags need to be tweaked on some architectures
+#       feel free to update the Makefile for your architecture and send a pull request or issue
+message(STATUS "CMAKE_SYSTEM_PROCESSOR: ${CMAKE_SYSTEM_PROCESSOR}")
+if (NOT MSVC)
+    if (LLAMA_STATIC)
+        add_link_options(-static)
+        if (MINGW)
+            add_link_options(-static-libgcc -static-libstdc++)
+        endif()
+    endif()
+    if (LLAMA_GPROF)
+        add_compile_options(-pg)
+    endif()
+    if (LLAMA_NATIVE)
+        add_compile_options(-march=native)
+    endif()
+endif()

-add_executable(llama
-    main.cpp
-    utils.cpp
-    utils.h)
+if (${CMAKE_SYSTEM_PROCESSOR} MATCHES "arm" OR ${CMAKE_SYSTEM_PROCESSOR} MATCHES "aarch64")
+    message(STATUS "ARM detected")
+    if (MSVC)
+        # TODO: arm msvc?
+    else()
+        if (${CMAKE_SYSTEM_PROCESSOR} MATCHES "aarch64")
+            add_compile_options(-mcpu=native)
+        endif()
+        # TODO: armv6,7,8 version specific flags
+    endif()
+elseif (${CMAKE_SYSTEM_PROCESSOR} MATCHES "^(x86_64|i686|AMD64)$")
+    message(STATUS "x86 detected")
+    if (MSVC)
+        if (LLAMA_AVX2)
+            add_compile_options(/arch:AVX2)
+        elseif (LLAMA_AVX)
+            add_compile_options(/arch:AVX)
+        endif()
+    else()
+        add_compile_options(-mf16c)
+        if (LLAMA_FMA)
+            add_compile_options(-mfma)
+        endif()
+        if (LLAMA_AVX)
+            add_compile_options(-mavx)
+        endif()
+        if (LLAMA_AVX2)
+            add_compile_options(-mavx2)
+        endif()
+    endif()
+else()
+    # TODO: support PowerPC
+    message(STATUS "Unknown architecture")
+endif()

-add_executable(quantize
-    quantize.cpp
-    utils.cpp
-    utils.h)

-add_library(ggml
-    ggml.c
-    ggml.h)
+#
+# Build library
+#

-target_compile_definitions(ggml PUBLIC ${LLAMA_EXTRA_FLAGS})
-target_compile_definitions(llama PUBLIC ${LLAMA_EXTRA_FLAGS})
-target_compile_definitions(quantize PUBLIC ${LLAMA_EXTRA_FLAGS})
+add_executable(llama main.cpp)
+
+add_executable(quantize quantize.cpp)
+
+add_library(utils OBJECT
+            utils.cpp
+            utils.h)
+
+target_include_directories(utils PUBLIC .)
+target_compile_features(utils PUBLIC cxx_std_11) # don't bump
+
+add_library(ggml OBJECT
+            ggml.c
+            ggml.h)

-target_link_libraries(ggml PRIVATE ${LLAMA_EXTRA_LIBS})
 target_include_directories(ggml PUBLIC .)
-target_link_libraries(quantize PRIVATE ggml)
-target_link_libraries(llama PRIVATE ggml)
-target_link_libraries(ggml PRIVATE Threads::Threads)
+target_compile_features(ggml PUBLIC c_std_11) # don't bump
+
+#
+# Linking
+#
+
+target_link_libraries(ggml PRIVATE Threads::Threads ${LLAMA_EXTRA_LIBS})
+target_link_libraries(llama PRIVATE ggml utils)
+target_link_libraries(quantize PRIVATE ggml utils)
+
+#
+# programs, examples and tests
+#
+
+if (LLAMA_BUILD_TESTS AND NOT CMAKE_JS_VERSION)
+    enable_testing()
+    add_subdirectory(tests)
+endif ()
+
+#if (LLAMA_BUILD_EXAMPLES)
+#    add_subdirectory(examples)
+#endif()
--- a/63
+++ b/63
@@ -17,7 +17,7 @@ CXXV := $(shell $(CXX) --version | head -n 1)
 # ref: https://github.com/ggerganov/whisper.cpp/issues/66#issuecomment-1282546789
 ifeq ($(UNAME_S),Darwin)
 	ifneq ($(UNAME_P),arm)
-		SYSCTL_M := $(shell sysctl -n hw.optional.arm64)
+		SYSCTL_M := $(shell sysctl -n hw.optional.arm64 2>/dev/null)
 		ifeq ($(SYSCTL_M),1)
 			# UNAME_P := arm
 			# UNAME_M := arm64
@@ -30,6 +30,7 @@ endif
 # Compile flags
 #

+# keep standard at C11 and C++11
 CFLAGS   = -I.              -O3 -DNDEBUG -std=c11   -fPIC
 CXXFLAGS = -I. -I./examples -O3 -DNDEBUG -std=c++11 -fPIC
 LDFLAGS  =
@@ -52,6 +53,10 @@ ifeq ($(UNAME_S),NetBSD)
 	CFLAGS   += -pthread
 	CXXFLAGS += -pthread
 endif
+ifeq ($(UNAME_S),OpenBSD)
+	CFLAGS   += -pthread
+	CXXFLAGS += -pthread
+endif
 ifeq ($(UNAME_S),Haiku)
 	CFLAGS   += -pthread
 	CXXFLAGS += -pthread
@@ -95,30 +100,59 @@ ifeq ($(UNAME_M),$(filter $(UNAME_M),x86_64 i686))
 		ifneq (,$(findstring sse3,$(SSE3_M)))
 			CFLAGS += -msse3
 		endif
+		AVX512F_M := $(shell grep "avx512f " /proc/cpuinfo)
+		ifneq (,$(findstring avx512f,$(AVX512F_M)))
+			CFLAGS += -mavx512f
+		endif
+		AVX512BW_M := $(shell grep "avx512bw " /proc/cpuinfo)
+		ifneq (,$(findstring avx512bw,$(AVX512BW_M)))
+			CFLAGS += -mavx512bw
+		endif
+		AVX512DQ_M := $(shell grep "avx512dq " /proc/cpuinfo)
+		ifneq (,$(findstring avx512dq,$(AVX512DQ_M)))
+			CFLAGS += -mavx512dq
+		endif
+		AVX512VL_M := $(shell grep "avx512vl " /proc/cpuinfo)
+		ifneq (,$(findstring avx512vl,$(AVX512VL_M)))
+			CFLAGS += -mavx512vl
+		endif
+		AVX512CD_M := $(shell grep "avx512cd " /proc/cpuinfo)
+		ifneq (,$(findstring avx512cd,$(AVX512CD_M)))
+			CFLAGS += -mavx512cd
+		endif
+		AVX512ER_M := $(shell grep "avx512er " /proc/cpuinfo)
+		ifneq (,$(findstring avx512er,$(AVX512ER_M)))
+			CFLAGS += -mavx512er
+		endif
+		AVX512IFMA_M := $(shell grep "avx512ifma " /proc/cpuinfo)
+		ifneq (,$(findstring avx512ifma,$(AVX512IFMA_M)))
+			CFLAGS += -mavx512ifma
+		endif
+		AVX512PF_M := $(shell grep "avx512pf " /proc/cpuinfo)
+		ifneq (,$(findstring avx512pf,$(AVX512PF_M)))
+			CFLAGS += -mavx512pf
+		endif
 	else ifeq ($(UNAME_S),Haiku)
-		AVX1_M := $(shell sysinfo -cpu | grep "AVX ")
-		ifneq (,$(findstring avx,$(AVX1_M)))
+		AVX1_M := $(shell sysinfo -cpu | grep -w "AVX")
+		ifneq (,$(findstring AVX,$(AVX1_M)))
 			CFLAGS += -mavx
 		endif
-		AVX2_M := $(shell sysinfo -cpu | grep "AVX2 ")
-		ifneq (,$(findstring avx2,$(AVX2_M)))
+		AVX2_M := $(shell sysinfo -cpu | grep -w "AVX2")
+		ifneq (,$(findstring AVX2,$(AVX2_M)))
 			CFLAGS += -mavx2
 		endif
-		FMA_M := $(shell sysinfo -cpu | grep "FMA ")
-		ifneq (,$(findstring fma,$(FMA_M)))
+		FMA_M := $(shell sysinfo -cpu | grep -w "FMA")
+		ifneq (,$(findstring FMA,$(FMA_M)))
 			CFLAGS += -mfma
 		endif
-		F16C_M := $(shell sysinfo -cpu | grep "F16C ")
-		ifneq (,$(findstring f16c,$(F16C_M)))
+		F16C_M := $(shell sysinfo -cpu | grep -w "F16C")
+		ifneq (,$(findstring F16C,$(F16C_M)))
 			CFLAGS += -mf16c
 		endif
 	else
 		CFLAGS += -mfma -mf16c -mavx -mavx2
 	endif
 endif
-ifeq ($(UNAME_M),amd64)
-	CFLAGS += -mavx -mavx2 -mfma -mf16c
-endif
 ifneq ($(filter ppc64%,$(UNAME_M)),)
 	POWER9_M := $(shell grep "POWER9" /proc/cpuinfo)
 	ifneq (,$(findstring POWER9,$(POWER9_M)))
@@ -130,7 +164,8 @@ ifneq ($(filter ppc64%,$(UNAME_M)),)
 	endif
 endif
 ifndef LLAMA_NO_ACCELERATE
-	# Mac M1 - include Accelerate framework
+	# Mac M1 - include Accelerate framework.
+	# `-framework Accelerate` works on Mac Intel as well, with negliable performance boost (as of the predict time).
 	ifeq ($(UNAME_S),Darwin)
 		CFLAGS  += -DGGML_USE_ACCELERATE
 		LDFLAGS += -framework Accelerate
@@ -193,7 +228,7 @@ clean:

 main: main.cpp ggml.o utils.o
 	$(CXX) $(CXXFLAGS) main.cpp ggml.o utils.o -o main $(LDFLAGS)
-	./main -h
+	@echo "\x1b[36mrun ./main -h for help\x1b[0m"

 quantize: quantize.cpp ggml.o utils.o
 	$(CXX) $(CXXFLAGS) quantize.cpp ggml.o utils.o -o quantize $(LDFLAGS)
--- a/README.md
+++ b/README.md
@@ -5,6 +5,15 @@

 Inference of [LLaMA](https://arxiv.org/abs/2302.13971) model in pure C/C++

+---
+
+**TEMPORARY NOTICE:**
+Big code change incoming: https://github.com/ggerganov/llama.cpp/pull/370
+
+Do not merge stuff until we merge this. Probably merge will happen on March 22 ~6:00am UTC
+
+---
+
 **Hot topics:**

 - [Added Alpaca support](https://github.com/ggerganov/llama.cpp#instruction-mode-with-alpaca)
@@ -147,7 +156,7 @@ python3 -m pip install torch numpy sentencepiece
 python3 convert-pth-to-ggml.py models/7B/ 1

 # quantize the model to 4-bits
-./quantize.sh 7B
+python3 quantize.py 7B

 # run the inference
 ./main -m ./models/7B/ggml-model-q4_0.bin -n 128
@@ -175,10 +184,15 @@ If you want a more ChatGPT-like experience, you can run in interactive mode by p
 In this mode, you can always interrupt generation by pressing Ctrl+C and enter one or more lines of text which will be converted into tokens and appended to the current context. You can also specify a *reverse prompt* with the parameter `-r "reverse prompt string"`. This will result in user input being prompted whenever the exact tokens of the reverse prompt string are encountered in the generation. A typical use is to use a prompt which makes LLaMa emulate a chat between multiple users, say Alice and Bob, and pass `-r "Alice:"`.

 Here is an example few-shot interaction, invoked with the command
-```
-./main -m ./models/13B/ggml-model-q4_0.bin -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt

+```bash
+# default arguments using 7B model
+./chat.sh
+
+# custom arguments using 13B model
+./main -m ./models/13B/ggml-model-q4_0.bin -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
 ```
+
 Note the use of `--color` to distinguish between user input and generated text.

 ![image](https://user-images.githubusercontent.com/1991296/224575029-2af3c7dc-5a65-4f64-a6bb-517a532aea38.png)
@@ -189,11 +203,10 @@ First, download the `ggml` Alpaca model into the `./models` folder:

 ```
 # use one of these
-# NOTE: these are copied from the alpaca.cpp repo - not sure how long these will work
 # TODO: add a script to simplify the download
-curl -o ggml-alpaca-7b-q4.bin -C - https://gateway.estuary.tech/gw/ipfs/QmQ1bf2BTnYxq73MFJWu1B7bQ2UD6qG7D7YDCxhTndVkPC
-curl -o ggml-alpaca-7b-q4.bin -C - https://ipfs.io/ipfs/QmQ1bf2BTnYxq73MFJWu1B7bQ2UD6qG7D7YDCxhTndVkPC
-curl -o ggml-alpaca-7b-q4.bin -C - https://cloudflare-ipfs.com/ipfs/QmQ1bf2BTnYxq73MFJWu1B7bQ2UD6qG7D7YDCxhTndVkPC
+curl -o ./models/ggml-alpaca-7b-q4.bin -C - https://gateway.estuary.tech/gw/ipfs/QmUp1UGeQFDqJKvtjbSYPBiZZKRjLp8shVP9hT8ZB9Ynv1
+curl -o ./models/ggml-alpaca-7b-q4.bin -C - https://ipfs.io/ipfs/QmUp1UGeQFDqJKvtjbSYPBiZZKRjLp8shVP9hT8ZB9Ynv1
+curl -o ./models/ggml-alpaca-7b-q4.bin -C - https://cloudflare-ipfs.com/ipfs/QmUp1UGeQFDqJKvtjbSYPBiZZKRjLp8shVP9hT8ZB9Ynv1
 ```

 Now run the `main` tool like this:
@@ -216,7 +229,7 @@ Sample run:
 There 26 letters in the English Alphabet
 > What is the most common way of transportation in Amsterdam?
 The majority (54%) are using public transit. This includes buses, trams and metros with over 100 lines throughout the city which make it very accessible for tourists to navigate around town as well as locals who commute by tram or metro on a daily basis
-> List 5 words that start with "ca".                                                                       
+> List 5 words that start with "ca".
 cadaver, cauliflower, cabbage (vegetable), catalpa (tree) and Cailleach.
 > 
 ```
--- a/alpaca.sh
+++ b/alpaca.sh
@@ -0,0 +1,6 @@
+#!/bin/bash
+#
+# Temporary script - will be removed in the future
+#
+
+./main -m ./models/ggml-alpaca-7b-q4.bin --color -f ./prompts/alpaca.txt -ins --top_k 10000 --temp 0.2 --repeat_penalty 1 -t 7
--- a/chat.sh
+++ b/chat.sh
@@ -0,0 +1,6 @@
+#!/bin/bash
+#
+# Temporary script - will be removed in the future
+#
+
+./main -m ./models/7B/ggml-model-q4_0.bin -n 256 --repeat_penalty 1.0 --color -i -r "User:" -f prompts/chat-with-bob.txt
--- a/convert-gptq-to-ggml.py
+++ b/convert-gptq-to-ggml.py
@@ -0,0 +1,172 @@
+# Convert a GPTQ quantized LLaMA model to a ggml compatible file
+# Based on: https://github.com/qwopqwop200/GPTQ-for-LLaMa
+#
+import os
+import re
+import sys
+import json
+import struct
+import numpy as np
+import torch
+from sentencepiece import SentencePieceProcessor
+
+if len(sys.argv) != 4:
+    print("Usage: convert-gptq-to-ggml.py llamaXXb-4bit.pt tokenizer.model out.bin\n")
+    sys.exit(1)
+
+fname_model = sys.argv[1]
+fname_tokenizer = sys.argv[2]
+dir_out = sys.argv[3]
+
+model = torch.load(fname_model, map_location="cpu")
+
+n_vocab, n_embd = model['model.embed_tokens.weight'].shape
+n_layer = 1 + max(int(m.group(1)) for name in model
+                  if (m := re.match(r'model\.layers\.([0-9]+)', name)))
+
+# hardcoded:
+n_mult = 256
+n_head = {32: 32, 40: 40, 60: 52, 80: 64}[n_layer]
+
+tokenizer = SentencePieceProcessor(fname_tokenizer)
+
+assert tokenizer.vocab_size() == n_vocab
+
+fname_out = sys.argv[3]
+
+fout = open(fname_out, "wb")
+
+fout.write(struct.pack("i", 0x67676d6c)) # magic: ggml in hex
+fout.write(struct.pack("i", n_vocab))
+fout.write(struct.pack("i", n_embd))
+fout.write(struct.pack("i", n_mult))
+fout.write(struct.pack("i", n_head))
+fout.write(struct.pack("i", n_layer))
+fout.write(struct.pack("i", n_embd // n_head)) # rot (obsolete)
+fout.write(struct.pack("i", 4))
+
+
+# This loop unchanged from convert-pth-to-ggml.py:
+for i in range(tokenizer.vocab_size()):
+    if tokenizer.is_unknown(i):
+        # "<unk>" token (translated as ??)
+        text = " \u2047 ".encode("utf-8")
+        fout.write(struct.pack("i", len(text)))
+        fout.write(text)
+    elif tokenizer.is_control(i):
+        # "<s>"/"</s>" tokens
+        fout.write(struct.pack("i", 0))
+    elif tokenizer.is_byte(i):
+        # "<U+XX>" tokens (which may be invalid UTF-8)
+        piece = tokenizer.id_to_piece(i)
+        if len(piece) != 6:
+            print("Invalid token: " + piece)
+            sys.exit(1)
+        byte_value = int(piece[3:-1], 16)
+        fout.write(struct.pack("i", 1))
+        fout.write(struct.pack("B", byte_value))
+    else:
+        # normal token. Uses U+2581 (LOWER ONE EIGHTH BLOCK) to represent spaces.
+        text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
+        fout.write(struct.pack("i", len(text)))
+        fout.write(text)
+
+def write_header(shape, dst_name, ftype_cur):
+    sname = dst_name.encode('utf-8')
+    fout.write(struct.pack("iii", len(shape), len(sname), ftype_cur))
+    fout.write(struct.pack("i" * len(shape), *shape[::-1]))
+    fout.write(sname)
+
+def convert_non_q4(src_name, dst_name):
+    v = model[src_name]
+    shape = v.shape
+    print("Processing non-Q4 variable: " + src_name + " with shape: ", shape, " and type: ", v.dtype)
+    if len(shape) == 1:
+        print("  Converting to float32")
+        v = v.to(torch.float32)
+
+    ftype_cur = {torch.float16: 1, torch.float32: 0}[v.dtype]
+
+    # header
+    write_header(shape, dst_name, ftype_cur)
+
+    # data
+    v.numpy().tofile(fout)
+
+def convert_q4(src_name, dst_name, permute=False):
+    zeros = model[f"{src_name}.zeros"].numpy()
+    scales = model[f"{src_name}.scales"].numpy()
+    bias = model[f"{src_name}.bias"].numpy()
+    qweight = model[f"{src_name}.qweight"].numpy().T # transpose
+
+    # Q4_1 does not support bias; good thing the bias is always all zeros.
+    assert not np.any(bias)
+
+    # Each int32 item is actually 8 int4 items packed together, and it's transposed.
+    shape = (qweight.shape[0], qweight.shape[1] * 8)
+
+    print("Processing Q4 variable: " + src_name + " with shape: ", shape)
+
+    # The output format has the int4 weights in groups of 32 rather than 8.
+    # It looks like this:
+    # For each row:
+    #   For each group of 32 columns:
+    #     - addend (float32, 4 bytes)
+    #     - scale (float32, 4 bytes)
+    #     - weights (int4 * 32, 16 bytes)
+    # Note that in the input, the scales and addends are shared between all
+    # the columns in a row, so we end up wasting quite a bit of memory with
+    # repeated scales and addends.
+
+    addends = -zeros # flip sign
+
+    # Since the output format is mixed between integers and floats, we have
+    # to hackily view the floats as int32s just so numpy will let us
+    # concatenate them.
+    addends_view = addends.view(dtype=np.int32)
+    scales_view = scales.view(dtype=np.int32)
+
+    # Split into groups of 4 columns (i.e. 32 columns of quantized data):
+    grouped = qweight.reshape([qweight.shape[0], qweight.shape[1] // 4, 4])
+
+    # Repeat addends and scales:
+    addends_rep = np.atleast_3d(addends_view).repeat(grouped.shape[1], axis=1)
+    scales_rep = np.atleast_3d(scales_view).repeat(grouped.shape[1], axis=1)
+
+    blob = np.concatenate([scales_rep, addends_rep, grouped], axis=2, casting='no')
+
+    if permute:
+        # Permute some rows to undo the permutation done by convert_llama_weights_to_hf.py.
+        # This can be done after the above conversion because it doesn't affect column order/layout.
+        blob = (blob.reshape(n_head, 2, shape[0] // n_head // 2, *blob.shape[1:])
+                    .swapaxes(1, 2)
+                    .reshape(blob.shape))
+
+    # header
+    write_header(shape, dst_name, 3) # ftype = Q4_1
+
+    # data
+    blob.tofile(fout)
+
+convert_non_q4("model.embed_tokens.weight", "tok_embeddings.weight")
+convert_non_q4("model.norm.weight", "norm.weight")
+convert_non_q4("lm_head.weight", "output.weight")
+
+for i in range(n_layer):
+    convert_q4(f"model.layers.{i}.self_attn.q_proj", f"layers.{i}.attention.wq.weight", permute=True)
+    convert_q4(f"model.layers.{i}.self_attn.k_proj", f"layers.{i}.attention.wk.weight", permute=True)
+    convert_q4(f"model.layers.{i}.self_attn.v_proj", f"layers.{i}.attention.wv.weight")
+    convert_q4(f"model.layers.{i}.self_attn.o_proj", f"layers.{i}.attention.wo.weight")
+
+    convert_q4(f"model.layers.{i}.mlp.gate_proj", f"layers.{i}.feed_forward.w1.weight")
+    convert_q4(f"model.layers.{i}.mlp.down_proj", f"layers.{i}.feed_forward.w2.weight")
+    convert_q4(f"model.layers.{i}.mlp.up_proj",   f"layers.{i}.feed_forward.w3.weight")
+
+    convert_non_q4(f"model.layers.{i}.input_layernorm.weight", f"layers.{i}.attention_norm.weight")
+    convert_non_q4(f"model.layers.{i}.post_attention_layernorm.weight", f"layers.{i}.ffn_norm.weight")
+
+
+fout.close()
+
+print("Done. Output file: " + fname_out)
+print("")
--- a/convert-pth-to-ggml.py
+++ b/convert-pth-to-ggml.py
@@ -10,25 +10,26 @@
 #   - Name (char[name_length])
 #   - Data (float[n_dims])
 #
-# By default, the bigger matrices are converted to 16-bit floats.
-# This can be disabled by adding the "use-f32" CLI argument.
-#
 # At the start of the ggml file we write the model parameters
 # and vocabulary.
 #
+
 import argparse
+import os
 import sys
 import json
 import struct
 import numpy as np
 import torch
+
 from sentencepiece import SentencePieceProcessor

 def parse_args():

    parser = argparse.ArgumentParser(description='Convert a LLaMA model checkpoint to a ggml compatible file')
-    parser.add_argument('dir_model', help='directory containing the model checkpoint')
-    parser.add_argument('ftype', type=int, choices=[0, 1], default=1, help='file type (0: float32, 1: float16)')
+    parser.add_argument('dir_model',  help='directory containing the model checkpoint')
+    parser.add_argument('ftype',      help='file type (0: float32, 1: float16)', type=int, choices=[0, 1], default=1)
+    parser.add_argument('vocab_only', help='only write vocab to file', type=int, default=0, nargs='?')
    return parser.parse_args()

 def get_n_parts(dim):
@@ -44,8 +45,14 @@ def get_n_parts(dim):

 def load_hparams_and_tokenizer(dir_model):

+    # `dir_model` is something like `models/7B` or `models/7B/`.
+    # "tokenizer.model" is expected under model's parent dir.
+    # When `dir_model` is a symlink, f"{dir_model}/../tokenizer.model" would not be found.
+    # Let's use the model's parent dir directly.
+    model_parent_dir = os.path.dirname(os.path.normpath(dir_model))
+
    fname_hparams = f"{dir_model}/params.json"
-    fname_tokenizer = f"{dir_model}/../tokenizer.model"
+    fname_tokenizer = f"{model_parent_dir}/tokenizer.model"

    with open(fname_hparams, "r") as f:
        hparams = json.load(f)
@@ -60,7 +67,8 @@ def write_header(fout, hparams, ftype):

    keys = ["vocab_size", "dim", "multiple_of", "n_heads", "n_layers"]
    values = [
-        0x67676d6c,  # magic: ggml in hex
+        0x67676d66,  # magic: ggmf in hex
+        1, # file version
        *[hparams[key] for key in keys],
        hparams["dim"] // hparams["n_heads"],  # rot (obsolete)
        ftype
@@ -85,6 +93,7 @@ def write_tokens(fout, tokenizer):
            text = tokenizer.id_to_piece(i).replace("\u2581", " ").encode("utf-8")
        fout.write(struct.pack("i", len(text)))
        fout.write(text)
+        fout.write(struct.pack("f", tokenizer.get_score(i)))

 def process_and_write_variables(fout, model, ftype):

@@ -125,6 +134,29 @@ def main():
    ftype_str = ["f32", "f16"]

    hparams, tokenizer = load_hparams_and_tokenizer(dir_model)
+
+    print(args)
+
+    # if only writing vocab to file
+    if args.vocab_only:
+
+        fname_model = f"{dir_model}/consolidated.00.pth"
+        fname_out = f"{dir_model}/ggml-vocab.bin"
+
+        print(f"Extracting only the vocab from '{fname_model}'\n")
+
+        model = torch.load(fname_model, map_location="cpu")
+
+        with open(fname_out, "wb") as fout:
+            fout.write(struct.pack("i", hparams["vocab_size"]))
+            write_tokens(fout, tokenizer)
+
+        del model
+
+        print(f"Done. Output file: {fname_out}\n")
+
+        return
+
    n_parts = get_n_parts(hparams["dim"])

    for p in range(n_parts):
@@ -142,6 +174,7 @@ def main():
            process_and_write_variables(fout, model, ftype)

        del model
+
        print(f"Done. Output file: {fname_out}, (part {p})\n")

 if __name__ == "__main__":
--- a/examples/chatLLaMa
+++ b/examples/chatLLaMa
@@ -0,0 +1,53 @@
+#!/bin/bash
+
+cd "$(dirname "$0")/.." || exit
+
+MODEL="${MODEL:-./models/13B/ggml-model-q4_0.bin}"
+USER_NAME="${USER_NAME:-User}"
+AI_NAME="${AI_NAME:-ChatLLaMa}"
+
+# Adjust to the number of CPU cores you want to use.
+N_THREAD="${N_THREAD:-8}"
+# Number of tokens to predict (made it larger than default because we want a long interaction)
+N_PREDICTS="${N_PREDICTS:-2048}"
+
+# Note: you can also override the generation options by specifying them on the command line:
+# For example, override the context size by doing: ./chatLLaMa --ctx_size 1024
+GEN_OPTIONS="${GEN_OPTIONS:---ctx_size 2048 --temp 0.7 --top_k 40 --top_p 0.5 --repeat_last_n 256 --repeat_penalty 1.17647}"
+
+# shellcheck disable=SC2086 # Intended splitting of GEN_OPTIONS
+./main $GEN_OPTIONS \
+  --model "$MODEL" \
+  --threads "$N_THREAD" \
+  --n_predict "$N_PREDICTS" \
+  --color --interactive \
+  --reverse-prompt "${USER_NAME}:" \
+  --prompt "
+Text transcript of a never ending dialog, where ${USER_NAME} interacts with an AI assistant named ${AI_NAME}.
+${AI_NAME} is helpful, kind, honest, friendly, good at writing and never fails to answer ${USER_NAME}’s requests immediately and with details and precision.
+There are no annotations like (30 seconds passed...) or (to himself), just what ${USER_NAME} and ${AI_NAME} say aloud to each other.
+The dialog lasts for years, the entirety of it is shared below. It's 10000 pages long.
+The transcript only includes text, it does not include markup like HTML and Markdown.
+
+$USER_NAME: Hello, $AI_NAME!
+$AI_NAME: Hello $USER_NAME! How may I help you today?
+$USER_NAME: What time is it?
+$AI_NAME: It is $(date +%H:%M).
+$USER_NAME: What year is it?
+$AI_NAME: We are in $(date +%Y).
+$USER_NAME: Please tell me the largest city in Europe.
+$AI_NAME: The largest city in Europe is Moscow, the capital of Russia.
+$USER_NAME: What can you tell me about Moscow?
+$AI_NAME: Moscow, on the Moskva River in western Russia, is the nation’s cosmopolitan capital. In its historic core is the Kremlin, a complex that’s home to the president and tsarist treasures in the Armoury. Outside its walls is Red Square, Russia’s symbolic center.
+$USER_NAME: What is a cat?
+$AI_NAME: A cat is a domestic species of small carnivorous mammal. It is the only domesticated species in the family Felidae.
+$USER_NAME: How do I pass command line arguments to a Node.js program?
+$AI_NAME: The arguments are stored in process.argv.
+
+    argv[0] is the path to the Node. js executable.
+    argv[1] is the path to the script file.
+    argv[2] is the first argument passed to the script.
+    argv[3] is the second argument passed to the script and so on.
+$USER_NAME: Name a color.
+$AI_NAME: Blue
+$USER_NAME:" "$@"
--- a/flake.nix
+++ b/flake.nix
@@ -34,6 +34,7 @@
            cat ${./convert-pth-to-ggml.py} >> $out/bin/convert-pth-to-ggml
            chmod +x $out/bin/convert-pth-to-ggml
          '';
+          meta.mainProgram = "llama";
        };
        devShells.default = pkgs.mkShell {
          packages = with pkgs; [
--- a/ggml.c
+++ b/ggml.c
@@ -2,7 +2,7 @@

 #if defined(_MSC_VER) || defined(__MINGW32__)
 #include <malloc.h> // using malloc.h with MSC/MINGW
-#elif !defined(__FreeBSD__) && !defined(__NetBSD__)
+#elif !defined(__FreeBSD__) && !defined(__NetBSD__) && !defined(__OpenBSD__)
 #include <alloca.h>
 #endif

@@ -361,7 +361,7 @@ static const size_t CACHE_LINE_SIZE_F32 = CACHE_LINE_SIZE/sizeof(float);

 // AVX routines provided by GH user Const-me
 // ref: https://github.com/ggerganov/ggml/pull/27#issuecomment-1464934600
-#if __AVX2__
+#if __AVX2__ || __AVX512F__
 // Unpack 32 4-bit fields into 32 bytes
 // The output vector contains 32 bytes, each one in [ 0 .. 15 ] interval
 static inline __m256i bytesFromNibbles( const uint8_t* rsi )
@@ -397,7 +397,6 @@ static inline __m128i packNibbles( __m256i bytes )
 }
 #endif

-
 // method 5
 // blocks of QK elements
 // represented with a single float (delta) and QK/2 8-bit ints (i.e QK 4-bit signed integer factors)
@@ -1262,6 +1261,47 @@ inline static void ggml_vec_dot_f32(const int n, float * restrict s, const float
    *s = sumf;
 }

+#if __AVX512F__ && QK == 32
+static inline __m512 dot_q4_0_oneblock_avx512(
+    __m512 acc,
+    const uint8_t * pd0,
+    const uint8_t * pd1,
+    const uint8_t * pb0,
+    const uint8_t * pb1,
+    size_t bs,
+    int i
+) {
+    const float * d0_0 = (const float *) (pd0 + i*bs);
+    const float * d1_0 = (const float *) (pd1 + i*bs);
+
+    const uint8_t * restrict p0 = pb0 + (i+0)*bs;
+    const uint8_t * restrict p1 = pb1 + (i+0)*bs;
+
+    // Compute combined scale for the block
+    float scaleScalar = d0_0[0] * d1_0[0];
+    __m512 scale = _mm512_set1_ps( scaleScalar );
+
+    __m256i bx = bytesFromNibbles( p0 );
+    __m256i by = bytesFromNibbles( p1 );
+
+    // Now we have a vector with bytes in [ 0 .. 15 ] interval. Offset them into [ -8 .. +7 ] interval.
+    const __m256i off = _mm256_set1_epi8( 8 );
+    bx = _mm256_sub_epi8( bx, off );
+    by = _mm256_sub_epi8( by, off );
+
+    // Sign-extend 16 signed bytes into int16_t
+    __m512i x32 = _mm512_cvtepi8_epi16( bx );
+    __m512i y32 = _mm512_cvtepi8_epi16( by );
+    // Compute products of int16_t integers, add pairwise
+    __m512i i64 = _mm512_madd_epi16( x32, y32 );
+
+    // Convert int32_t to float
+    __m512 p = _mm512_cvtepi32_ps( i64 );
+    // Apply the scale, and accumulate
+    return _mm512_fmadd_ps( scale, p, acc );
+}
+#endif
+
 inline static void ggml_vec_dot_f16(const int n, float * restrict s, ggml_fp16_t * restrict x, ggml_fp16_t * restrict y) {
    ggml_float sumf = 0.0;

@@ -1417,6 +1457,40 @@ inline static void ggml_vec_dot_q4_0(const int n, float * restrict s, const void
 #else
 #error "not implemented for QK"
 #endif
+#elif defined(__AVX512F__)
+
+#if QK == 32
+    // Initialize accumulator with zeros
+    __m512 acc0 = _mm512_setzero_ps();
+    __m512 acc1 = _mm512_setzero_ps();
+
+    const int superblock_size = 8;
+    const int superblock_count = nb / superblock_size;
+    const int remainder = nb % superblock_size;
+
+    for (int superblock_ix = 0; superblock_ix < superblock_count; superblock_ix += 1) {
+        int i = superblock_ix * superblock_size;
+
+        acc0 = dot_q4_0_oneblock_avx512( acc0, pd0, pd1, pb0, pb1, bs, i+0 );
+        acc1 = dot_q4_0_oneblock_avx512( acc1, pd0, pd1, pb0, pb1, bs, i+1 );
+        acc0 = dot_q4_0_oneblock_avx512( acc0, pd0, pd1, pb0, pb1, bs, i+2 );
+        acc1 = dot_q4_0_oneblock_avx512( acc1, pd0, pd1, pb0, pb1, bs, i+3 );
+        acc0 = dot_q4_0_oneblock_avx512( acc0, pd0, pd1, pb0, pb1, bs, i+4 );
+        acc1 = dot_q4_0_oneblock_avx512( acc1, pd0, pd1, pb0, pb1, bs, i+5 );
+        acc0 = dot_q4_0_oneblock_avx512( acc0, pd0, pd1, pb0, pb1, bs, i+6 );
+        acc1 = dot_q4_0_oneblock_avx512( acc1, pd0, pd1, pb0, pb1, bs, i+7 );
+    }
+
+    // Remainders
+    for (int i = superblock_count * superblock_size; i < nb; ++i) {
+        acc0 = dot_q4_0_oneblock_avx512( acc0, pd0, pd1, pb0, pb1, bs, i );
+    }
+
+    // Horizontal sum of all lanes of the accumulator
+    sumf = _mm512_reduce_add_ps( acc0 ) + _mm512_reduce_add_ps( acc1 );
+#else
+#error "not implemented for QK"
+#endif
 #elif defined(__AVX2__)
 #if QK == 32
    const size_t countBlocks = nb;
@@ -1928,7 +2002,7 @@ inline static void ggml_vec_mad_q4_1(const int n, float * restrict y, void * res
    const size_t bs = 2*sizeof(float) + QK/2;

    const uint8_t * restrict pd = ((const uint8_t *)x + 0*bs);
-    const uint8_t * restrict pm = ((const uint8_t *)x + 0*bs +   sizeof(float)); 
+    const uint8_t * restrict pm = ((const uint8_t *)x + 0*bs +   sizeof(float));
    const uint8_t * restrict pb = ((const uint8_t *)x + 0*bs + 2*sizeof(float));

    for (int i = 0; i < nb; i++) {
--- a/main.cpp
+++ b/main.cpp
@@ -3,11 +3,12 @@
 #include "utils.h"

 #include <cassert>
+#include <cinttypes>
 #include <cmath>
 #include <cstdio>
 #include <cstring>
 #include <fstream>
-#include <map>
+#include <iostream>
 #include <string>
 #include <vector>

@@ -18,6 +19,13 @@
 #include <signal.h>
 #endif

+#if defined (_WIN32)
+#pragma comment(lib,"kernel32.lib")
+extern "C" __declspec(dllimport) void* __stdcall GetStdHandle(unsigned long nStdHandle);
+extern "C" __declspec(dllimport) int __stdcall GetConsoleMode(void* hConsoleHandle, unsigned long* lpMode);
+extern "C" __declspec(dllimport) int __stdcall SetConsoleMode(void* hConsoleHandle, unsigned long dwMode);
+#endif
+
 #define ANSI_COLOR_RED     "\x1b[31m"
 #define ANSI_COLOR_GREEN   "\x1b[32m"
 #define ANSI_COLOR_YELLOW  "\x1b[33m"
@@ -27,8 +35,40 @@
 #define ANSI_COLOR_RESET   "\x1b[0m"
 #define ANSI_BOLD          "\x1b[1m"

+/* Keep track of current color of output, and emit ANSI code if it changes. */
+enum console_state {
+    CONSOLE_STATE_DEFAULT=0,
+    CONSOLE_STATE_PROMPT,
+    CONSOLE_STATE_USER_INPUT
+}; 
+
+static console_state con_st = CONSOLE_STATE_DEFAULT;
+static bool con_use_color = false;
+
+void set_console_state(console_state new_st)
+{
+    if (!con_use_color) return;
+    // only emit color code if state changed
+    if (new_st != con_st) {
+        con_st = new_st;
+        switch(con_st) {
+        case CONSOLE_STATE_DEFAULT:
+            printf(ANSI_COLOR_RESET);
+            return;
+        case CONSOLE_STATE_PROMPT:
+            printf(ANSI_COLOR_YELLOW);
+            return;
+        case CONSOLE_STATE_USER_INPUT:
+            printf(ANSI_BOLD ANSI_COLOR_GREEN);
+            return;
+        }
+    }
+}
+
+static const int EOS_TOKEN_ID = 2;
+
 // determine number of model parts based on the dimension
-static const std::map<int, int> LLAMA_N_PARTS = {
+static const std::unordered_map<int, int> LLAMA_N_PARTS = {
    { 4096, 1 },
    { 5120, 2 },
    { 6656, 4 },
@@ -82,11 +122,12 @@ struct llama_model {

    //
    struct ggml_context * ctx;
-    std::map<std::string, struct ggml_tensor *> tensors;
+    std::unordered_map<std::string, struct ggml_tensor *> tensors;
 };

 // load the model's weights from a file
-bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab & vocab, int n_ctx, ggml_type memory_type = GGML_TYPE_F32) {
+
+bool llama_model_load(const std::string & fname, llama_model & model, llama_vocab & vocab, int n_ctx, int n_parts, ggml_type memory_type = GGML_TYPE_F32) {
    fprintf(stderr, "%s: loading model from '%s' - please wait ...\n", __func__, fname.c_str());

    std::vector<char> f_buf(1024*1024);
@@ -102,14 +143,27 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
    {
        uint32_t magic;
        fin.read((char *) &magic, sizeof(magic));
-        if (magic != 0x67676d6c) {
+        if (magic == FILE_MAGIC_UNVERSIONED) {
+            fprintf(stderr, "%s: invalid model file '%s' (too old, regenerate your model files!)\n",
+                    __func__, fname.c_str());
+            return false;
+        }
+        if (magic != FILE_MAGIC) {
            fprintf(stderr, "%s: invalid model file '%s' (bad magic)\n", __func__, fname.c_str());
            return false;
        }
+
+        uint32_t format_version;
+        fin.read((char *) &format_version, sizeof(format_version));
+
+        if (format_version != FILE_VERSION) {
+            fprintf(stderr, "%s: invalid model file '%s' (unsupported format version %" PRIu32 ", expected %d)\n",
+                    __func__, fname.c_str(), format_version, FILE_VERSION);
+            return false;
+        }
    }

    int n_ff = 0;
-    int n_parts = 0;

    // load hparams
    {
@@ -127,7 +181,16 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
        hparams.n_ctx = n_ctx;

        n_ff = ((2*(4*hparams.n_embd)/3 + hparams.n_mult - 1)/hparams.n_mult)*hparams.n_mult;
-        n_parts = LLAMA_N_PARTS.at(hparams.n_embd);
+
+        if (n_parts < 1) {
+            n_parts = LLAMA_N_PARTS.at(hparams.n_embd);
+        }
+
+        // temp warning to tell the user to use "--n_parts"
+        if (hparams.f16 == 4 && n_parts != 1) {
+            fprintf(stderr, "%s: GPTQ model detected - are you sure n_parts should be %d? we normally expect it to be 1\n", __func__, n_parts);
+            fprintf(stderr, "%s: use '--n_parts 1' if necessary\n", __func__);
+        }

        fprintf(stderr, "%s: n_vocab = %d\n", __func__, hparams.n_vocab);
        fprintf(stderr, "%s: n_ctx   = %d\n", __func__, hparams.n_ctx);
@@ -144,30 +207,43 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
    // load vocab
    {
        std::string word;
+        vocab.id_to_token.resize(model.hparams.n_vocab);
+        std::vector<char> tmp(64);
+
        for (int i = 0; i < model.hparams.n_vocab; i++) {
            uint32_t len;
            fin.read((char *) &len, sizeof(len));

            word.resize(len);
-            fin.read((char *) word.data(), len);
+            if (len > 0) {
+                tmp.resize(len);
+                fin.read(tmp.data(), len);
+                word.assign(tmp.data(), len);
+            } else {
+                word.clear();
+            }
+
+            float score;
+            fin.read((char *) &score, sizeof(score));

            vocab.token_to_id[word] = i;
-            vocab.id_to_token[i] = word;

-            //if (i < 30000) {
-            //    fprintf(stderr, "%s: vocab[%d] = '%s'\n", __func__, i, word.c_str());
-            //}
+            auto &tok_score = vocab.id_to_token[i];
+            tok_score.tok = word;
+            tok_score.score = score;
        }
    }

    // for the big tensors, we have the option to store the data in 16-bit floats or quantized
    // in order to save memory and also to speed up the computation
-    ggml_type wtype = GGML_TYPE_COUNT;
+    // wtype is for per-layer weights, while vtype is for other weights
+    ggml_type wtype, vtype;
    switch (model.hparams.f16) {
-        case 0: wtype = GGML_TYPE_F32;  break;
-        case 1: wtype = GGML_TYPE_F16;  break;
-        case 2: wtype = GGML_TYPE_Q4_0; break;
-        case 3: wtype = GGML_TYPE_Q4_1; break;
+        case 0: wtype = vtype = GGML_TYPE_F32;  break;
+        case 1: wtype = vtype = GGML_TYPE_F16;  break;
+        case 2: wtype = vtype = GGML_TYPE_Q4_0; break;
+        case 3: wtype = vtype = GGML_TYPE_Q4_1; break;
+        case 4: wtype = GGML_TYPE_Q4_1; vtype = GGML_TYPE_F16; break;
        default:
                {
                    fprintf(stderr, "%s: invalid model file '%s' (bad f16 value %d)\n",
@@ -188,11 +264,11 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab
        const int n_ctx   = hparams.n_ctx;
        const int n_vocab = hparams.n_vocab;

-        ctx_size += n_embd*n_vocab*ggml_type_sizef(wtype); // tok_embeddings
+        ctx_size += n_embd*n_vocab*ggml_type_sizef(vtype); // tok_embeddings

        ctx_size += n_embd*ggml_type_sizef(GGML_TYPE_F32); // norm

-        ctx_size += n_embd*n_vocab*ggml_type_sizef(wtype); // output
+        ctx_size += n_embd*n_vocab*ggml_type_sizef(vtype); // output

        ctx_size += n_layer*(n_embd*ggml_type_sizef(GGML_TYPE_F32)); // attention_norm

@@ -239,10 +315,10 @@ bool llama_model_load(const std::string & fname, llama_model & model, gpt_vocab

        model.layers.resize(n_layer);

-        model.tok_embeddings = ggml_new_tensor_2d(ctx, wtype, n_embd, n_vocab);
+        model.tok_embeddings = ggml_new_tensor_2d(ctx, vtype, n_embd, n_vocab);

        model.norm   = ggml_new_tensor_1d(ctx, GGML_TYPE_F32, n_embd);
-        model.output = ggml_new_tensor_2d(ctx, wtype,         n_embd, n_vocab);
+        model.output = ggml_new_tensor_2d(ctx, vtype,         n_embd, n_vocab);

        // map by name
        model.tensors["tok_embeddings.weight"] = model.tok_embeddings;
@@ -522,9 +598,10 @@ bool llama_eval(
        const llama_model & model,
        const int n_threads,
        const int n_past,
-        const std::vector<gpt_vocab::id> & embd_inp,
-              std::vector<float>         & embd_w,
-              size_t                     & mem_per_token) {
+        const std::vector<llama_vocab::id> & embd_inp,
+              std::vector<float>           & embd_w,
+              size_t                       & mem_per_token,
+              bool return_all_logits = false) {
    const int N = embd_inp.size();

    const auto & hparams = model.hparams;
@@ -542,7 +619,7 @@ bool llama_eval(
    static void * buf = malloc(buf_size);

    if (mem_per_token > 0 && mem_per_token*N > buf_size) {
-        const size_t buf_size_new = 1.1*(mem_per_token*N); // add 10% to account for ggml object overhead
+        const size_t buf_size_new = 1.3*(mem_per_token*N); // add 30% to account for ggml object overhead
        //fprintf(stderr, "\n%s: reallocating buffer from %zu to %zu bytes\n", __func__, buf_size, buf_size_new);

        // reallocate
@@ -728,9 +805,14 @@ bool llama_eval(
    //embd_w.resize(n_vocab*N);
    //memcpy(embd_w.data(), ggml_get_data(inpL), sizeof(float)*n_vocab*N);

-    // return result for just the last token
-    embd_w.resize(n_vocab);
-    memcpy(embd_w.data(), (float *) ggml_get_data(inpL) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
+    if (return_all_logits) {
+        embd_w.resize(n_vocab * N);
+        memcpy(embd_w.data(), (float *) ggml_get_data(inpL), sizeof(float)*n_vocab*N);
+    } else {
+        // return result for just the last token
+        embd_w.resize(n_vocab);
+        memcpy(embd_w.data(), (float *) ggml_get_data(inpL) + (n_vocab*(N-1)), sizeof(float)*n_vocab);
+    }

    if (mem_per_token == 0) {
        mem_per_token = ggml_used_mem(ctx0)/N;
@@ -742,11 +824,81 @@ bool llama_eval(
    return true;
 }

+std::vector<double> softmax(const std::vector<float>& logits) {
+    std::vector<double> probs(logits.size());
+    float max_logit = logits[0];
+    for (float v : logits) max_logit = std::max(max_logit, v);
+    double sum_exp = 0.0;
+    for (size_t i = 0; i < logits.size(); i++) {
+        // Subtract the maximum logit value from the current logit value for numerical stability
+        float logit = logits[i] - max_logit;
+        double exp_logit = std::exp(logit);
+        sum_exp += exp_logit;
+        probs[i] = exp_logit;
+    }
+    for (size_t i = 0; i < probs.size(); i++) probs[i] /= sum_exp;
+    return probs;
+}
+
+void perplexity(const llama_vocab &vocab, const llama_model &model, const gpt_params &params, size_t mem_per_token) {
+    // Download: https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-2-raw-v1.zip?ref=salesforce-research
+    // Run `./main --perplexity -m models/7B/ggml-model-q4_0.bin -f wiki.test.raw`
+    // Output: `perplexity: 13.5106 [114/114]`
+    std::vector<llama_vocab::id> tokens = ::llama_tokenize(vocab, params.prompt, true);
+
+    int count = 0;
+    double nll = 0.0;
+    int seq_count = tokens.size() / params.n_ctx;
+    printf("Calculating perplexity over %d chunks\n", seq_count);
+    for (int i = 0; i < seq_count; ++i) {
+        int start = i * params.n_ctx;
+        int end = start + params.n_ctx - 1;
+        std::vector<llama_vocab::id> embd(tokens.begin() + start, tokens.begin() + end);
+        std::vector<float> logits;
+        auto start_t = std::chrono::high_resolution_clock::now();
+        if (!llama_eval(model, params.n_threads, 0, embd, logits, mem_per_token, true)) {
+            fprintf(stderr, "Failed to predict\n");
+            return;
+        }
+        auto end_t = std::chrono::high_resolution_clock::now();
+        if (i == 0) {
+            double seconds = std::chrono::duration<double>(end_t - start_t).count();
+            printf("%.2f seconds per pass - ETA %.2f hours\n", seconds, (seconds * seq_count) / (60.0*60.0));
+        }
+        // We get the logits for all the tokens in the context window (params.n_ctx)
+        // from llama_eval above.  Now, based on https://huggingface.co/docs/transformers/perplexity,
+        // calculate the perplexity over the last half the window (so the model always has
+        // some context to predict the token).
+        //
+        // We rely on the fact that attention in the forward pass only looks at previous
+        // tokens here, so the logits returned for each token are an accurate representation
+        // of what the model would have predicted at that point.
+        //
+        // Example, we have a context window of 512, we will compute perplexity for each of the
+        // last 256 tokens.  Then, we split the input up into context window size chunks to
+        // process the entire prompt.
+        for (int j = params.n_ctx / 2; j < params.n_ctx - 1; ++j) {
+            // Calculate probability of next token, given the previous ones.
+            int n_vocab = model.hparams.n_vocab;
+            std::vector<float> tok_logits(
+                logits.begin() + j * n_vocab,
+                logits.begin() + (j + 1) * n_vocab);
+            double prob = softmax(tok_logits)[tokens[start + j + 1]];
+            nll += -std::log(prob);
+            ++count;
+        }
+        // perplexity is e^(average negative log-likelihood)
+        printf("[%d]%.4lf,", i + 1, std::exp(nll / count));
+        fflush(stdout);
+    }
+    printf("\n");
+}
+
 static bool is_interacting = false;

 #if defined (__unix__) || (defined (__APPLE__) && defined (__MACH__)) || defined (_WIN32)
 void sigint_handler(int signo) {
-    printf(ANSI_COLOR_RESET);
+    set_console_state(CONSOLE_STATE_DEFAULT);
    printf("\n"); // this also force flush stdout.
    if (signo == SIGINT) {
        if (!is_interacting) {
@@ -801,23 +953,27 @@ int main(int argc, char ** argv) {
    fprintf(stderr, "%s: seed = %d\n", __func__, params.seed);

    std::mt19937 rng(params.seed);
-    if (params.prompt.empty()) {
+    if (params.random_prompt) {
        params.prompt = gpt_random_prompt(rng);
    }

+    // save choice to use color for later
+    // (note for later: this is a slightly awkward choice)
+    con_use_color = params.use_color;
+
 //    params.prompt = R"(// this function checks if the number n is prime
 //bool is_prime(int n) {)";

    int64_t t_load_us = 0;

-    gpt_vocab vocab;
+    llama_vocab vocab;
    llama_model model;

    // load the model
    {
        const ggml_type memory_type = params.memory_f16 ? GGML_TYPE_F16 : GGML_TYPE_F32;
        const int64_t t_start_us = ggml_time_us();
-        if (!llama_model_load(params.model, model, vocab, params.n_ctx, memory_type)) {
+        if (!llama_model_load(params.model, model, vocab, params.n_ctx, params.n_parts, memory_type)) {
            fprintf(stderr, "%s: failed to load model from '%s'\n", __func__, params.model.c_str());
            return 1;
        }
@@ -832,35 +988,41 @@ int main(int argc, char ** argv) {
                params.n_threads, std::thread::hardware_concurrency(), llama_print_system_info());
    }

+    std::vector<float> logits;
+
+    // determine the required inference memory per token:
+    size_t mem_per_token = 0;
+    llama_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
+
+    if (params.perplexity) {
+        perplexity(vocab, model, params, mem_per_token);
+        exit(0);
+    }
+
    int n_past = 0;

    int64_t t_sample_us  = 0;
    int64_t t_predict_us = 0;

-    std::vector<float> logits;
-
    // Add a space in front of the first character to match OG llama tokenizer behavior
    params.prompt.insert(0, 1, ' ');
    // tokenize the prompt
-    std::vector<gpt_vocab::id> embd_inp = ::llama_tokenize(vocab, params.prompt, true);
+    std::vector<llama_vocab::id> embd_inp = ::llama_tokenize(vocab, params.prompt, true);

    params.n_predict = std::min(params.n_predict, model.hparams.n_ctx - (int) embd_inp.size());

    // prefix & suffix for instruct mode
-    const std::vector<gpt_vocab::id> inp_pfx = ::llama_tokenize(vocab, "\n\n### Instruction:\n\n", true);
-    const std::vector<gpt_vocab::id> inp_sfx = ::llama_tokenize(vocab, "\n\n### Response:\n\n", false);
+    const std::vector<llama_vocab::id> inp_pfx = ::llama_tokenize(vocab, "\n\n### Instruction:\n\n", true);
+    const std::vector<llama_vocab::id> inp_sfx = ::llama_tokenize(vocab, "\n\n### Response:\n\n", false);

    // in instruct mode, we inject a prefix and a suffix to each input by the user
    if (params.instruct) {
        params.interactive = true;
-        params.antiprompt = "### Instruction:\n\n";
+        params.antiprompt.push_back("### Instruction:\n\n");
    }

-    // tokenize the reverse prompt
-    std::vector<gpt_vocab::id> antiprompt_inp = ::llama_tokenize(vocab, params.antiprompt, false);
-
    // enable interactive mode if reverse prompt is specified
-    if (!antiprompt_inp.empty()) {
+    if (params.antiprompt.size() != 0) {
        params.interactive = true;
    }

@@ -868,7 +1030,7 @@ int main(int argc, char ** argv) {
    fprintf(stderr, "%s: prompt: '%s'\n", __func__, params.prompt.c_str());
    fprintf(stderr, "%s: number of tokens in prompt = %zu\n", __func__, embd_inp.size());
    for (int i = 0; i < (int) embd_inp.size(); i++) {
-        fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], vocab.id_to_token.at(embd_inp[i]).c_str());
+        fprintf(stderr, "%6d -> '%s'\n", embd_inp[i], vocab.id_to_token.at(embd_inp[i]).tok.c_str());
    }
    fprintf(stderr, "\n");
    if (params.interactive) {
@@ -884,26 +1046,19 @@ int main(int argc, char ** argv) {

        fprintf(stderr, "%s: interactive mode on.\n", __func__);

-        if (antiprompt_inp.size()) {
-            fprintf(stderr, "%s: reverse prompt: '%s'\n", __func__, params.antiprompt.c_str());
-            fprintf(stderr, "%s: number of tokens in reverse prompt = %zu\n", __func__, antiprompt_inp.size());
-            for (int i = 0; i < (int) antiprompt_inp.size(); i++) {
-                fprintf(stderr, "%6d -> '%s'\n", antiprompt_inp[i], vocab.id_to_token.at(antiprompt_inp[i]).c_str());
+        if(params.antiprompt.size()) {
+            for (auto antiprompt : params.antiprompt) {
+                fprintf(stderr, "Reverse prompt: '%s'\n", antiprompt.c_str());
            }
-            fprintf(stderr, "\n");
        }
    }
    fprintf(stderr, "sampling parameters: temp = %f, top_k = %d, top_p = %f, repeat_last_n = %i, repeat_penalty = %f\n", params.temp, params.top_k, params.top_p, params.repeat_last_n, params.repeat_penalty);
    fprintf(stderr, "\n\n");

-    std::vector<gpt_vocab::id> embd;
-
-    // determine the required inference memory per token:
-    size_t mem_per_token = 0;
-    llama_eval(model, params.n_threads, 0, { 0, 1, 2, 3 }, logits, mem_per_token);
+    std::vector<llama_vocab::id> embd;

    int last_n_size = params.repeat_last_n;
-    std::vector<gpt_vocab::id> last_n_tokens(last_n_size);
+    std::vector<llama_vocab::id> last_n_tokens(last_n_size);
    std::fill(last_n_tokens.begin(), last_n_tokens.end(), 0);

    if (params.interactive) {
@@ -921,10 +1076,18 @@ int main(int argc, char ** argv) {

    int remaining_tokens = params.n_predict;

-    // set the color for the prompt which will be output initially
-    if (params.use_color) {
-        printf(ANSI_COLOR_YELLOW);
+#if defined (_WIN32)
+  if (params.use_color) {
+        // Enable ANSI colors on Windows 10+
+        unsigned long dwMode = 0;
+        void* hConOut = GetStdHandle((unsigned long)-11); // STD_OUTPUT_HANDLE (-11)
+        if (hConOut && hConOut != (void*)-1 && GetConsoleMode(hConOut, &dwMode) && !(dwMode & 0x4)) {
+            SetConsoleMode(hConOut, dwMode | 0x4); // ENABLE_VIRTUAL_TERMINAL_PROCESSING (0x4)
+        }
    }
+#endif
+    // the first thing we will do is to output the prompt, so set color accordingly
+    set_console_state(CONSOLE_STATE_PROMPT);

    while (remaining_tokens > 0 || params.interactive) {
        // predict
@@ -942,7 +1105,7 @@ int main(int argc, char ** argv) {
        n_past += embd.size();
        embd.clear();

-        if (embd_inp.size() <= input_consumed) {
+        if ((int) embd_inp.size() <= input_consumed) {
            // out of user input, sample next token
            const float top_k = params.top_k;
            const float top_p = params.top_p;
@@ -951,11 +1114,16 @@ int main(int argc, char ** argv) {

            const int n_vocab = model.hparams.n_vocab;

-            gpt_vocab::id id = 0;
+            llama_vocab::id id = 0;

            {
                const int64_t t_start_sample_us = ggml_time_us();

+                if (params.ignore_eos) {
+                    // set the logit of the eos token to zero to avoid sampling it
+                    logits[logits.size() - n_vocab + EOS_TOKEN_ID] = 0;
+                }
+
                id = llama_sample_top_p_top_k(vocab, logits.data() + (logits.size() - n_vocab), last_n_tokens, repeat_penalty, top_k, top_p, temp, rng);

                last_n_tokens.erase(last_n_tokens.begin());
@@ -974,7 +1142,7 @@ int main(int argc, char ** argv) {
            --remaining_tokens;
        } else {
            // some user input remains from prompt or interaction, forward it to processing
-            while (embd_inp.size() > input_consumed) {
+            while ((int) embd_inp.size() > input_consumed) {
                embd.push_back(embd_inp[input_consumed]);
                last_n_tokens.erase(last_n_tokens.begin());
                last_n_tokens.push_back(embd_inp[input_consumed]);
@@ -983,30 +1151,40 @@ int main(int argc, char ** argv) {
                    break;
                }
            }
-
-            // reset color to default if we there is no pending user input
-            if (!input_noecho && params.use_color && (int) embd_inp.size() == input_consumed) {
-                printf(ANSI_COLOR_RESET);
-            }
        }

        // display text
        if (!input_noecho) {
            for (auto id : embd) {
-                printf("%s", vocab.id_to_token[id].c_str());
+                printf("%s", vocab.id_to_token[id].tok.c_str());
            }
            fflush(stdout);
        }
+        // reset color to default if we there is no pending user input
+        if (!input_noecho && (int)embd_inp.size() == input_consumed) {
+            set_console_state(CONSOLE_STATE_DEFAULT);
+        }

        // in interactive mode, and not currently processing queued inputs;
        // check if we should prompt the user for more
-        if (params.interactive && embd_inp.size() <= input_consumed) {
+        if (params.interactive && (int) embd_inp.size() <= input_consumed) {
            // check for reverse prompt
-            if (antiprompt_inp.size() && std::equal(antiprompt_inp.rbegin(), antiprompt_inp.rend(), last_n_tokens.rbegin())) {
-                // reverse prompt found
-                is_interacting = true;
+            std::string last_output;
+            for (auto id : last_n_tokens) {
+                last_output += vocab.id_to_token[id].tok;
+            }
+
+            // Check if each of the reverse prompts appears at the end of the output.
+            for (std::string antiprompt : params.antiprompt) {
+                if (last_output.find(antiprompt.c_str(), last_output.length() - antiprompt.length(), antiprompt.length()) != std::string::npos) {
+                    is_interacting = true;
+                    break;
+                }
            }
            if (is_interacting) {
+                // potentially set color to indicate we are taking user input
+                set_console_state(CONSOLE_STATE_USER_INPUT);
+
                if (params.instruct) {
                    input_consumed = embd_inp.size();
                    embd_inp.insert(embd_inp.end(), inp_pfx.begin(), inp_pfx.end());
@@ -1014,48 +1192,38 @@ int main(int argc, char ** argv) {
                    printf("\n> ");
                }

-                // currently being interactive
+                std::string buffer;
+                std::string line;
                bool another_line = true;
-                while (another_line) {
-                    fflush(stdout);
-                    char buf[256] = {0};
-                    int n_read;
-                    if (params.use_color) printf(ANSI_BOLD ANSI_COLOR_GREEN);
-                    if (scanf("%255[^\n]%n%*c", buf, &n_read) <= 0) {
-                        // presumable empty line, consume the newline
-                        std::ignore = scanf("%*c");
-                        n_read=0;
-                    }
-                    if (params.use_color) printf(ANSI_COLOR_RESET);
-
-                    if (n_read > 0 && buf[n_read-1]=='\\') {
-                        another_line = true;
-                        buf[n_read-1] = '\n';
-                        buf[n_read] = 0;
-                    } else {
+                do {
+                    std::getline(std::cin, line);
+                    if (line.empty() || line.back() != '\\') {
                        another_line = false;
-                        buf[n_read] = '\n';
-                        buf[n_read+1] = 0;
+                    } else {
+                        line.pop_back(); // Remove the continue character
                    }
+                    buffer += line + '\n'; // Append the line to the result
+                } while (another_line);

-                    std::vector<gpt_vocab::id> line_inp = ::llama_tokenize(vocab, buf, false);
-                    embd_inp.insert(embd_inp.end(), line_inp.begin(), line_inp.end());
+                // done taking input, reset color
+                set_console_state(CONSOLE_STATE_DEFAULT);

-                    if (params.instruct) {
-                        embd_inp.insert(embd_inp.end(), inp_sfx.begin(), inp_sfx.end());
-                    }
+                std::vector<llama_vocab::id> line_inp = ::llama_tokenize(vocab, buffer, false);
+                embd_inp.insert(embd_inp.end(), line_inp.begin(), line_inp.end());

-                    remaining_tokens -= line_inp.size();
-
-                    input_noecho = true; // do not echo this again
+                if (params.instruct) {
+                    embd_inp.insert(embd_inp.end(), inp_sfx.begin(), inp_sfx.end());
                }

-                is_interacting = false;
+                remaining_tokens -= line_inp.size();
+
+                input_noecho = true; // do not echo this again
            }
+            is_interacting = false;
        }

        // end of text token
-        if (embd.back() == 2) {
+        if (embd.back() == EOS_TOKEN_ID) {
            if (params.interactive) {
                is_interacting = true;
            } else {
@@ -1063,6 +1231,12 @@ int main(int argc, char ** argv) {
                break;
            }
        }
+
+        // In interactive mode, respect the maximum number of tokens and drop back to user input when reached.
+        if (params.interactive && remaining_tokens <= 0) {
+            remaining_tokens = params.n_predict;
+            is_interacting = true;
+        }
    }

 #if defined (_WIN32)
@@ -1083,9 +1257,7 @@ int main(int argc, char ** argv) {

    ggml_free(model.ctx);

-    if (params.use_color) {
-        printf(ANSI_COLOR_RESET);
-    }
+    set_console_state(CONSOLE_STATE_DEFAULT);

    return 0;
 }
--- a/models/ggml-vocab.bin
+++ b/models/ggml-vocab.bin
--- a/quantize.cpp
+++ b/quantize.cpp
@@ -3,11 +3,11 @@
 #include "utils.h"

 #include <cassert>
+#include <cinttypes>
 #include <cmath>
 #include <cstdio>
 #include <cstring>
 #include <fstream>
-#include <map>
 #include <string>
 #include <vector>
 #include <regex>
@@ -43,7 +43,7 @@ bool llama_model_quantize(const std::string & fname_inp, const std::string & fna
        return false;
    }

-    gpt_vocab vocab;
+    llama_vocab vocab;

    printf("%s: loading model from '%s'\n", __func__, fname_inp.c_str());

@@ -63,12 +63,28 @@ bool llama_model_quantize(const std::string & fname_inp, const std::string & fna
    {
        uint32_t magic;
        finp.read((char *) &magic, sizeof(magic));
-        if (magic != 0x67676d6c) {
+        if (magic == FILE_MAGIC_UNVERSIONED) {
+            fprintf(stderr, "%s: invalid model file '%s' (too old, regenerate your model files!)\n",
+                    __func__, fname_inp.c_str());
+            return false;
+        }
+        if (magic != FILE_MAGIC) {
            fprintf(stderr, "%s: invalid model file '%s' (bad magic)\n", __func__, fname_inp.c_str());
            return false;
        }

        fout.write((char *) &magic, sizeof(magic));
+
+        uint32_t format_version;
+        finp.read((char *) &format_version, sizeof(format_version));
+
+        if (format_version != FILE_VERSION) {
+            fprintf(stderr, "%s: invalid model file '%s' (unsupported format version %" PRIu32 ", expected %d)\n",
+                    __func__, fname_inp.c_str(), format_version, FILE_VERSION);
+            return false;
+        }
+
+        fout.write((char *) &format_version, sizeof(format_version));
    }

    llama_hparams hparams;
@@ -113,6 +129,7 @@ bool llama_model_quantize(const std::string & fname_inp, const std::string & fna
        }

        std::string word;
+        vocab.id_to_token.resize(n_vocab);
        for (int i = 0; i < n_vocab; i++) {
            uint32_t len;
            finp.read ((char *) &len, sizeof(len));
@@ -122,8 +139,15 @@ bool llama_model_quantize(const std::string & fname_inp, const std::string & fna
            finp.read ((char *) word.data(), len);
            fout.write((char *) word.data(), len);

+            float score;
+            finp.read ((char *) &score, sizeof(score));
+            fout.write((char *) &score, sizeof(score));
+
            vocab.token_to_id[word] = i;
-            vocab.id_to_token[i] = word;
+
+            auto &tok_score = vocab.id_to_token[i];
+            tok_score.tok = word;
+            tok_score.score = score;
        }
    }

--- a/quantize.py
+++ b/quantize.py
@@ -0,0 +1,126 @@
+#!/usr/bin/env python3
+
+"""Script to execute the "quantize" script on a given set of models."""
+
+import subprocess
+import argparse
+import glob
+import sys
+import os
+
+
+def main():
+    """Update the quantize binary name depending on the platform and parse
+    the command line arguments and execute the script.
+    """
+
+    if "linux" in sys.platform or "darwin" in sys.platform:
+        quantize_script_binary = "quantize"
+
+    elif "win32" in sys.platform or "cygwin" in sys.platform:
+        quantize_script_binary = "quantize.exe"
+
+    else:
+        print("WARNING: Unknown platform. Assuming a UNIX-like OS.\n")
+        quantize_script_binary = "quantize"
+
+    parser = argparse.ArgumentParser(
+        prog='python3 quantize.py',
+        description='This script quantizes the given models by applying the '
+        f'"{quantize_script_binary}" script on them.'
+    )
+    parser.add_argument(
+        'models', nargs='+', choices=('7B', '13B', '30B', '65B'),
+        help='The models to quantize.'
+    )
+    parser.add_argument(
+        '-r', '--remove-16', action='store_true', dest='remove_f16',
+        help='Remove the f16 model after quantizing it.'
+    )
+    parser.add_argument(
+        '-m', '--models-path', dest='models_path',
+        default=os.path.join(os.getcwd(), "models"),
+        help='Specify the directory where the models are located.'
+    )
+    parser.add_argument(
+        '-q', '--quantize-script-path', dest='quantize_script_path',
+        default=os.path.join(os.getcwd(), quantize_script_binary),
+        help='Specify the path to the "quantize" script.'
+    )
+
+    # TODO: Revise this code
+    # parser.add_argument(
+    #     '-t', '--threads', dest='threads', type='int',
+    #     default=os.cpu_count(),
+    #     help='Specify the number of threads to use to quantize many models at '
+    #     'once. Defaults to os.cpu_count().'
+    # )
+
+    args = parser.parse_args()
+
+    if not os.path.isfile(args.quantize_script_path):
+        print(
+            f'The "{quantize_script_binary}" script was not found in the '
+            "current location.\nIf you want to use it from another location, "
+            "set the --quantize-script-path argument from the command line."
+        )
+        sys.exit(1)
+
+    for model in args.models:
+        # The model is separated in various parts
+        # (ggml-model-f16.bin, ggml-model-f16.bin.0, ggml-model-f16.bin.1...)
+        f16_model_path_base = os.path.join(
+            args.models_path, model, "ggml-model-f16.bin"
+        )
+
+        f16_model_parts_paths = map(
+            lambda filename: os.path.join(f16_model_path_base, filename),
+            glob.glob(f"{f16_model_path_base}*")
+        )
+
+        for f16_model_part_path in f16_model_parts_paths:
+            if not os.path.isfile(f16_model_part_path):
+                print(
+                    f"The f16 model {os.path.basename(f16_model_part_path)} "
+                    f"was not found in {args.models_path}{os.path.sep}{model}"
+                    ". If you want to use it from another location, set the "
+                    "--models-path argument from the command line."
+                )
+                sys.exit(1)
+
+            __run_quantize_script(
+                args.quantize_script_path, f16_model_part_path
+            )
+
+            if args.remove_f16:
+                os.remove(f16_model_part_path)
+
+
+# This was extracted to a top-level function for parallelization, if
+# implemented. See https://github.com/ggerganov/llama.cpp/pull/222/commits/f8db3d6cd91bf1a1342db9d29e3092bc12dd783c#r1140496406
+
+def __run_quantize_script(script_path, f16_model_part_path):
+    """Run the quantize script specifying the path to it and the path to the
+    f16 model to quantize.
+    """
+
+    new_quantized_model_path = f16_model_part_path.replace("f16", "q4_0")
+    subprocess.run(
+        [script_path, f16_model_part_path, new_quantized_model_path, "2"],
+        check=True
+    )
+
+
+if __name__ == "__main__":
+    try:
+        main()
+
+    except subprocess.CalledProcessError:
+        print("\nAn error ocurred while trying to quantize the models.")
+        sys.exit(1)
+
+    except KeyboardInterrupt:
+        sys.exit(0)
+
+    else:
+        print("\nSuccesfully quantized all models.")
--- a/quantize.sh
+++ b/quantize.sh
@@ -1,15 +0,0 @@
-#!/usr/bin/env bash
-
-if ! [[ "$1" =~ ^[0-9]{1,2}B$ ]]; then
-    echo
-    echo "Usage: quantize.sh 7B|13B|30B|65B [--remove-f16]"
-    echo
-    exit 1
-fi
-
-for i in `ls models/$1/ggml-model-f16.bin*`; do
-    ./quantize "$i" "${i/f16/q4_0}" 2
-    if [[ "$2" == "--remove-f16" ]]; then
-        rm "$i"
-    fi
-done
--- a/tests/CMakeLists.txt
+++ b/tests/CMakeLists.txt
@@ -0,0 +1,4 @@
+set(TEST_TARGET test-tokenizer-0)
+add_executable(${TEST_TARGET} ${TEST_TARGET}.cpp)
+target_link_libraries(${TEST_TARGET} PRIVATE utils)
+add_test(NAME ${TEST_TARGET} COMMAND $<TARGET_FILE:${TEST_TARGET}> ${CMAKE_CURRENT_SOURCE_DIR}/../models/ggml-vocab.bin)
--- a/tests/test-tokenizer-0.cpp
+++ b/tests/test-tokenizer-0.cpp
@@ -0,0 +1,69 @@
+#include "utils.h"
+
+#include <cstdio>
+#include <string>
+#include <map>
+
+static const std::map<std::string, std::vector<llama_vocab::id>> k_tests = {
+    { "Hello World",        { 1,  10994,   2787, }, },
+    { " Hello World",       { 1,  15043,   2787, }, },
+    { " Hello World!",      { 1,  15043,   2787,  29991, }, },
+    { " this is 🦙.cpp",    { 1,    445,    338,  29871,    243,    162,    169,    156,  29889,   8223, }, },
+    { "w048 7tuijk dsdfhu", { 1,  29893,  29900,  29946,  29947,  29871,  29955,   9161,  13535,  18031,   2176,   6905, }, },
+    { "нещо на Български",  { 1,    821,   4851,    665,   1386,  29713,   1305, }, },
+};
+
+int main(int argc, char **argv) {
+    if (argc < 2) {
+        fprintf(stderr, "Usage: %s <vocab-file>\n", argv[0]);
+        return 1;
+    }
+
+    const std::string fname = argv[1];
+
+    fprintf(stderr, "%s : reading vocab from: '%s'\n", __func__, fname.c_str());
+
+    llama_vocab vocab;
+
+    if (!llama_vocab_load(fname, vocab)) {
+        fprintf(stderr, "%s : failed to load vocab from: '%s'\n", __func__, fname.c_str());
+        return 1;
+    }
+
+    const int n_vocab = vocab.id_to_token.size();
+
+    if (n_vocab != 32000) {
+        fprintf(stderr, "%s : expected 32000 tokens, got %d\n", __func__, n_vocab);
+        return 2;
+    }
+
+    for (const auto & test_kv : k_tests) {
+        const auto res = llama_tokenize(vocab, test_kv.first, true);
+
+        bool correct = res.size() == test_kv.second.size();
+
+        for (int i = 0; i < (int) res.size() && correct; ++i) {
+            if (res[i] != test_kv.second[i]) {
+                correct = false;
+            }
+        }
+
+        if (!correct) {
+            fprintf(stderr, "%s : failed test: '%s'\n", __func__, test_kv.first.c_str());
+            fprintf(stderr, "%s : expected tokens: ", __func__);
+            for (const auto & t : test_kv.second) {
+                fprintf(stderr, "%6d, ", t);
+            }
+            fprintf(stderr, "\n");
+            fprintf(stderr, "%s : got tokens:      ", __func__);
+            for (const auto & t : res) {
+                fprintf(stderr, "%6d, ", t);
+            }
+            fprintf(stderr, "\n");
+
+            return 3;
+        }
+    }
+
+    return 0;
+}
--- a/utils.cpp
+++ b/utils.cpp
@@ -6,12 +6,13 @@
 #include <regex>
 #include <iostream>
 #include <iterator>
+#include <queue>
 #include <string>
 #include <math.h>

 #if defined(_MSC_VER) || defined(__MINGW32__)
 #include <malloc.h> // using malloc.h with MSC/MINGW
- #elif !defined(__FreeBSD__) && !defined(__NetBSD__)
+ #elif !defined(__FreeBSD__) && !defined(__NetBSD__) && !defined(__OpenBSD__)
 #include <alloca.h>
 #endif

@@ -70,10 +71,18 @@ bool gpt_params_parse(int argc, char ** argv, gpt_params & params) {
        } else if (arg == "--color") {
            params.use_color = true;
        } else if (arg == "-r" || arg == "--reverse-prompt") {
-            params.antiprompt = argv[++i];
+            params.antiprompt.push_back(argv[++i]);
+        } else if (arg == "--perplexity") {
+            params.perplexity = true;
+        } else if (arg == "--ignore-eos") {
+            params.ignore_eos = true;
+        } else if (arg == "--n_parts") {
+            params.n_parts = std::stoi(argv[++i]);
        } else if (arg == "-h" || arg == "--help") {
            gpt_print_usage(argc, argv, params);
            exit(0);
+        } else if (arg == "--random-prompt") {
+            params.random_prompt = true;
        } else {
            fprintf(stderr, "error: unknown argument: %s\n", arg.c_str());
            gpt_print_usage(argc, argv, params);
@@ -92,12 +101,14 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
    fprintf(stderr, "  -i, --interactive     run in interactive mode\n");
    fprintf(stderr, "  -ins, --instruct      run in instruction mode (use with Alpaca models)\n");
    fprintf(stderr, "  -r PROMPT, --reverse-prompt PROMPT\n");
-    fprintf(stderr, "                        in interactive mode, poll user input upon seeing PROMPT\n");
+    fprintf(stderr, "                        in interactive mode, poll user input upon seeing PROMPT (can be\n");
+    fprintf(stderr, "                        specified more than once for multiple prompts).\n");
    fprintf(stderr, "  --color               colorise output to distinguish prompt and user input from generations\n");
    fprintf(stderr, "  -s SEED, --seed SEED  RNG seed (default: -1)\n");
    fprintf(stderr, "  -t N, --threads N     number of threads to use during computation (default: %d)\n", params.n_threads);
    fprintf(stderr, "  -p PROMPT, --prompt PROMPT\n");
-    fprintf(stderr, "                        prompt to start generation with (default: random)\n");
+    fprintf(stderr, "                        prompt to start generation with (default: empty)\n");
+    fprintf(stderr, "  --random-prompt       start with a randomized prompt.\n");
    fprintf(stderr, "  -f FNAME, --file FNAME\n");
    fprintf(stderr, "                        prompt file to start generation.\n");
    fprintf(stderr, "  -n N, --n_predict N   number of tokens to predict (default: %d)\n", params.n_predict);
@@ -106,9 +117,12 @@ void gpt_print_usage(int /*argc*/, char ** argv, const gpt_params & params) {
    fprintf(stderr, "  --repeat_last_n N     last n tokens to consider for penalize (default: %d)\n", params.repeat_last_n);
    fprintf(stderr, "  --repeat_penalty N    penalize repeat sequence of tokens (default: %.1f)\n", params.repeat_penalty);
    fprintf(stderr, "  -c N, --ctx_size N    size of the prompt context (default: %d)\n", params.n_ctx);
+    fprintf(stderr, "  --ignore-eos          ignore end of stream token and continue generating\n");
    fprintf(stderr, "  --memory_f16          use f16 instead of f32 for memory key+value\n");
    fprintf(stderr, "  --temp N              temperature (default: %.1f)\n", params.temp);
+    fprintf(stderr, "  --n_parts N           number of model parts (default: -1 = determine from dimensions)\n");
    fprintf(stderr, "  -b N, --batch_size N  batch size for prompt processing (default: %d)\n", params.n_batch);
+    fprintf(stderr, "  --perplexity          compute perplexity over the prompt\n");
    fprintf(stderr, "  -m FNAME, --model FNAME\n");
    fprintf(stderr, "                        model path (default: %s)\n", params.model.c_str());
    fprintf(stderr, "\n");
@@ -141,8 +155,8 @@ void replace(std::string & str, const std::string & needle, const std::string &
    }
 }

-std::map<std::string, int32_t> json_parse(const std::string & fname) {
-    std::map<std::string, int32_t> result;
+std::unordered_map<std::string, int32_t> json_parse(const std::string & fname) {
+    std::unordered_map<std::string, int32_t> result;

    // read file into string
    std::string json;
@@ -232,151 +246,213 @@ std::map<std::string, int32_t> json_parse(const std::string & fname) {
    return result;
 }

-std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text) {
-    std::vector<std::string> words;
-
-    // first split the text into words
-    {
-        std::string str = text;
-        std::string pat = R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)";
-
-        std::regex re(pat);
-        std::smatch m;
-
-        while (std::regex_search(str, m, re)) {
-            for (auto x : m) {
-                words.push_back(x);
-            }
-            str = m.suffix();
-        }
-    }
-
-    // find the longest tokens that form the words:
-    std::vector<gpt_vocab::id> tokens;
-    for (const auto & word : words) {
-        if (word.size() == 0) continue;
-
-        int i = 0;
-        int n = word.size();
-        while (i < n) {
-            int j = n;
-            while (j > i) {
-                auto it = vocab.token_to_id.find(word.substr(i, j-i));
-                if (it != vocab.token_to_id.end()) {
-                    tokens.push_back(it->second);
-                    i = j;
-                    break;
-                }
-                --j;
-            }
-            if (i == n) {
-                break;
-            }
-            if (j == i) {
-                auto sub = word.substr(i, 1);
-                if (vocab.token_to_id.find(sub) != vocab.token_to_id.end()) {
-                    tokens.push_back(vocab.token_to_id.at(sub));
-                } else {
-                    fprintf(stderr, "%s: unknown token '%s'\n", __func__, sub.data());
-                }
-                ++i;
-            }
-        }
-    }
-
-    return tokens;
+static size_t utf8_len(char src) {
+    const size_t lookup[] = { 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 3, 4 };
+    uint8_t highbits = static_cast<uint8_t>(src) >> 4;
+    return lookup[highbits];
 }

-// TODO: Calculate this constant from the vocabulary
-#define MAX_TOKEN_LEN 18
-// SentencePiece implementation after https://guillaume-be.github.io/2020-05-30/sentence_piece
-std::vector<gpt_vocab::id> llama_tokenize(const gpt_vocab & vocab, const std::string & text, bool bos) {
-    std::vector<gpt_vocab::id> res;
-    std::vector<int> score;
-    std::vector<gpt_vocab::id> prev;
-    int len = text.length();
+struct llama_sp_symbol {
+    using index = int;
+    index prev;
+    index next;
+    const char * text;
+    size_t n;
+};

-    score.resize(len + 1);
-    prev.resize(len + 1);
+struct llama_sp_bigram {
+    struct comparator {
+        bool operator()(llama_sp_bigram & l, llama_sp_bigram & r) {
+            return (l.score < r.score) || (l.score == r.score && l.left > r.left);
+        }
+    };
+    using queue_storage = std::vector<llama_sp_bigram>;
+    using queue = std::priority_queue<llama_sp_bigram, queue_storage, comparator>;
+    llama_sp_symbol::index left;
+    llama_sp_symbol::index right;
+    float score;
+    size_t size;
+};

-    // Forward pass
-    for (int i = 0; i < len; i++) {
-        int max_len = std::min(len - i, MAX_TOKEN_LEN);
-        for (int sub_len = 1; sub_len <= max_len; sub_len++) {
-            auto sub = text.substr(i, sub_len);
-            auto token = vocab.token_to_id.find(sub);
-            if (token != vocab.token_to_id.end()) {
-                int token_score = sub.length() * sub.length();
-                int local_score = score[i] + token_score;
-                int next = i + sub_len;
-                if (score[next] < local_score) {
-                    score[next] = local_score;
-                    prev[next] = (*token).second;
+// original implementation:
+// https://github.com/ggerganov/llama.cpp/commit/074bea2eb1f1349a0118239c4152914aecaa1be4
+struct llama_tokenizer {
+    llama_tokenizer(const llama_vocab & vocab): vocab_(vocab) {}
+
+    void tokenize(const std::string & text, std::vector<llama_vocab::id> & output) {
+        // split string into utf8 chars
+        int index = 0;
+        size_t offs = 0;
+        while (offs < text.size()) {
+            llama_sp_symbol sym;
+            size_t char_len = std::min(text.size() - offs, utf8_len(text[offs]));
+            sym.text = text.c_str() + offs;
+            sym.n = char_len;
+            offs += char_len;
+            sym.prev = index - 1;
+            sym.next = offs == text.size() ? -1 : index + 1;
+            index++;
+            symbols_.emplace_back(std::move(sym));
+        }
+
+        // seed the work queue with all possible 2-character tokens.
+        for (size_t i = 1; i < symbols_.size(); ++i) {
+            try_add_bigram(i - 1, i);
+        }
+
+        // keep substituting the highest frequency pairs for as long as we can.
+        while (!work_queue_.empty()) {
+            auto bigram = work_queue_.top();
+            work_queue_.pop();
+
+            auto & left_sym = symbols_[bigram.left];
+            auto & right_sym = symbols_[bigram.right];
+
+            // if one of the symbols already got merged, skip it.
+            if (left_sym.n == 0 || right_sym.n == 0 ||
+                left_sym.n + right_sym.n != bigram.size) {
+                continue;
+            }
+
+            // merge the right sym into the left one
+            left_sym.n += right_sym.n;
+            right_sym.n = 0;
+
+            //printf("left = '%*s' size = %zu\n", (int) left_sym.n, left_sym.text, bigram.size);
+
+            // remove the right sym from the chain
+            left_sym.next = right_sym.next;
+            if (right_sym.next >= 0) {
+                symbols_[right_sym.next].prev = bigram.left;
+            }
+
+            // find more substitutions
+            try_add_bigram(left_sym.prev, bigram.left);
+            try_add_bigram(bigram.left, left_sym.next);
+        }
+
+        for (int i = 0; i != -1; i = symbols_[i].next) {
+            auto & symbol = symbols_[i];
+            auto token = vocab_.token_to_id.find(std::string(symbol.text, symbol.n));
+
+            if (token == vocab_.token_to_id.end()) {
+                // output any symbols that did not form tokens as bytes.
+                for (int j = 0; j < (int) symbol.n; ++j) {
+                    llama_vocab::id token_id = static_cast<uint8_t>(symbol.text[j]) + 3;
+                    output.push_back(token_id);
                }
+            } else {
+                output.push_back((*token).second);
            }
        }
    }

-    // Backward pass
-    int i = len;
-    while (i > 0) {
-        gpt_vocab::id token_id = prev[i];
-        if (token_id == 0) {
-	    // TODO: Return error or something more meaningful
-            printf("failed to tokenize string!\n");
-	    break;
+private:
+    void try_add_bigram(int left, int right) {
+        if (left == -1 || right == -1) {
+            return;
        }
-        res.push_back(token_id);
-        auto token = (*vocab.id_to_token.find(token_id)).second;
-        i -= token.length();
+
+        const std::string text = std::string(symbols_[left].text, symbols_[left].n + symbols_[right].n);
+        auto token = vocab_.token_to_id.find(text);
+
+        if (token == vocab_.token_to_id.end()) {
+            return;
+        }
+
+        if (static_cast<size_t>((*token).second) >= vocab_.id_to_token.size()) {
+            return;
+        }
+
+        const auto &tok_score = vocab_.id_to_token[(*token).second];
+
+        llama_sp_bigram bigram;
+        bigram.left = left;
+        bigram.right = right;
+        bigram.score = tok_score.score;
+        bigram.size = text.size();
+        work_queue_.push(bigram);
    }

-    if (bos) {
-        res.push_back(1); // TODO: replace with vocab.bos
+    const llama_vocab & vocab_;
+    std::vector<llama_sp_symbol> symbols_;
+    llama_sp_bigram::queue work_queue_;
+};
+
+// TODO: temporary code duplication with llama.cpp
+//       will resolve after #77 is merged
+bool llama_vocab_load(const std::string & fname, llama_vocab & vocab) {
+    std::ifstream fin(fname, std::ios::binary);
+    if (!fin.is_open()) {
+        return false;
    }

-    // Pieces are in reverse order so correct that
-    std::reverse(res.begin(), res.end());
+    int n_vocab = 0;
+    fin.read((char *) &n_vocab, sizeof(n_vocab));

-    return res;
-}
+    std::string word;
+    std::vector<char> tmp(64);

-bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab) {
-    printf("%s: loading vocab from '%s'\n", __func__, fname.c_str());
+    vocab.id_to_token.resize(n_vocab);

-    vocab.token_to_id = ::json_parse(fname);
+    for (int i = 0; i < n_vocab; i++) {
+        uint32_t len;
+        fin.read((char *) &len, sizeof(len));

-    for (const auto & kv : vocab.token_to_id) {
-        vocab.id_to_token[kv.second] = kv.first;
+        word.resize(len);
+        if (len > 0) {
+            tmp.resize(len);
+            fin.read(tmp.data(), len);
+            word.assign(tmp.data(), len);
+        } else {
+            word.clear();
+        }
+
+        float score;
+        fin.read((char *) &score, sizeof(score));
+
+        vocab.token_to_id[word] = i;
+
+        auto &tok_score = vocab.id_to_token[i];
+        tok_score.tok = word;
+        tok_score.score = score;
    }

-    printf("%s: vocab size = %d\n", __func__, (int) vocab.token_to_id.size());
-
-    // print the vocabulary
-    //for (auto kv : vocab.token_to_id) {
-    //    printf("'%s' -> %d\n", kv.first.data(), kv.second);
-    //}
-
    return true;
 }

+std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos) {
+    llama_tokenizer tokenizer(vocab);
+    std::vector<llama_vocab::id> output;

-void sample_top_k(std::vector<std::pair<double, gpt_vocab::id>> & logits_id, int top_k) {
+    if (text.size() == 0) {
+        return output;
+    }
+
+    if (bos) {
+        output.push_back(1);
+    }
+
+    tokenizer.tokenize(text, output);
+    return output;
+}
+
+void sample_top_k(std::vector<std::pair<double, llama_vocab::id>> & logits_id, int top_k) {
    // find the top K tokens
    std::partial_sort(
            logits_id.begin(),
            logits_id.begin() + top_k, logits_id.end(),
-            [](const std::pair<double, gpt_vocab::id> & a, const std::pair<double, gpt_vocab::id> & b) {
+            [](const std::pair<double, llama_vocab::id> & a, const std::pair<double, llama_vocab::id> & b) {
        return a.first > b.first;
    });

    logits_id.resize(top_k);
 }

-gpt_vocab::id llama_sample_top_p_top_k(
-        const gpt_vocab & vocab,
+llama_vocab::id llama_sample_top_p_top_k(
+        const llama_vocab & vocab,
        const float * logits,
-        std::vector<gpt_vocab::id> & last_n_tokens,
+        std::vector<llama_vocab::id> & last_n_tokens,
        double repeat_penalty,
        int top_k,
        double top_p,
@@ -384,7 +460,7 @@ gpt_vocab::id llama_sample_top_p_top_k(
        std::mt19937 & rng) {
    int n_logits = vocab.id_to_token.size();

-    std::vector<std::pair<double, gpt_vocab::id>> logits_id;
+    std::vector<std::pair<double, llama_vocab::id>> logits_id;
    logits_id.reserve(n_logits);

    {
@@ -527,7 +603,7 @@ size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t

    char * pdst = (char *) dst;

-    for (int j = 0; j < n; j += k) { 
+    for (int j = 0; j < n; j += k) {
        uint8_t * pd = (uint8_t *) (pdst + (j/k)*row_size + 0*bs);
        uint8_t * pm = (uint8_t *) (pdst + (j/k)*row_size + 0*bs +   sizeof(float));
        uint8_t * pb = (uint8_t *) (pdst + (j/k)*row_size + 0*bs + 2*sizeof(float));
@@ -550,7 +626,7 @@ size_t ggml_quantize_q4_1(float * src, void * dst, int n, int k, int qk, int64_t

                *(float *) pd = d;
                *(float *) pm = min;
-                pd += bs; 
+                pd += bs;
                pm += bs;

                for (int l = 0; l < qk; l += 2) {
--- a/utils.h
+++ b/utils.h
@@ -3,7 +3,7 @@
 #pragma once

 #include <string>
-#include <map>
+#include <unordered_map>
 #include <vector>
 #include <random>
 #include <thread>
@@ -13,29 +13,34 @@
 //

 struct gpt_params {
-    int32_t seed      = -1; // RNG seed
-    int32_t n_threads = std::min(4, (int32_t) std::thread::hardware_concurrency());
-    int32_t n_predict = 128; // new tokens to predict
+    int32_t seed          = -1;  // RNG seed
+    int32_t n_threads     = std::min(4, (int32_t) std::thread::hardware_concurrency());
+    int32_t n_predict     = 128; // new tokens to predict
    int32_t repeat_last_n = 64;  // last n tokens to penalize
-    int32_t n_ctx = 512; //context size
-    bool memory_f16 = false; // use f16 instead of f32 for memory kv
+    int32_t n_parts       = -1;  // amount of model parts (-1 = determine from model dimensions)
+    int32_t n_ctx         = 512; //context size

    // sampling parameters
    int32_t top_k = 40;
    float   top_p = 0.95f;
    float   temp  = 0.80f;
-    float   repeat_penalty  = 1.30f;
+    float   repeat_penalty  = 1.10f;

    int32_t n_batch = 8; // batch size for prompt processing

-    std::string model      = "models/lamma-7B/ggml-model.bin"; // model path
-    std::string prompt     = "";
-    std::string antiprompt = ""; // string upon seeing which more user input is prompted
+    std::string model  = "models/lamma-7B/ggml-model.bin"; // model path
+    std::string prompt = "";

-    bool use_color = false; // use color to distinguish generations and inputs
+    std::vector<std::string> antiprompt; // string upon seeing which more user input is prompted

-    bool interactive = false; // interactive mode
-    bool instruct    = false; // instruction mode (used for Alpaca models)
+    bool memory_f16        = false; // use f16 instead of f32 for memory kv
+    bool random_prompt     = false; // do not randomize prompt if none provided
+    bool use_color         = false; // use color to distinguish generations and inputs
+    bool interactive       = false; // interactive mode
+    bool interactive_start = false; // reverse prompt immediately
+    bool instruct          = false; // instruction mode (used for Alpaca models)
+    bool ignore_eos        = false; // do not stop generating after eos
+    bool perplexity        = false; // compute perplexity over the prompt
 };

 bool gpt_params_parse(int argc, char ** argv, gpt_params & params);
@@ -44,51 +49,52 @@ void gpt_print_usage(int argc, char ** argv, const gpt_params & params);

 std::string gpt_random_prompt(std::mt19937 & rng);

+//
+// Model file parsing
+//
+
+#define FILE_MAGIC_UNVERSIONED 0x67676d6c // pre-versioned files
+#define FILE_MAGIC 0x67676d66 // 'ggmf' in hex
+#define FILE_VERSION 1
+
 //
 // Vocab utils
 //

-struct gpt_vocab {
+struct llama_vocab {
    using id    = int32_t;
    using token = std::string;

-    std::map<token, id> token_to_id;
-    std::map<id, token> id_to_token;
+    struct token_score {
+        token tok;
+        float score;
+    };
+
+    std::unordered_map<token, id> token_to_id;
+    std::vector<token_score> id_to_token;
 };

 void replace(std::string & str, const std::string & needle, const std::string & replacement);

 // poor-man's JSON parsing
-std::map<std::string, int32_t> json_parse(const std::string & fname);
+std::unordered_map<std::string, int32_t> json_parse(const std::string & fname);

-// split text into tokens
-//
-// ref: https://github.com/openai/gpt-2/blob/a74da5d99abaaba920de8131d64da2862a8f213b/src/encoder.py#L53
-//
-// Regex (Python):
-// r"""'s|'t|'re|'ve|'m|'ll|'d| ?\p{L}+| ?\p{N}+| ?[^\s\p{L}\p{N}]+|\s+(?!\S)|\s+"""
-//
-// Regex (C++):
-// R"('s|'t|'re|'ve|'m|'ll|'d| ?[[:alpha:]]+| ?[[:digit:]]+| ?[^\s[:alpha:][:digit:]]+|\s+(?!\S)|\s+)"
-//
-std::vector<gpt_vocab::id> gpt_tokenize(const gpt_vocab & vocab, const std::string & text);
+// TODO: temporary until #77 is merged, need this now for some tokenizer tests
+bool llama_vocab_load(const std::string & fname, llama_vocab & vocab);

 // TODO: this is probably wrong, but I cannot figure out how this tokenizer works ..
 // ref: https://github.com/google/sentencepiece
-std::vector<gpt_vocab::id> llama_tokenize(const gpt_vocab & vocab, const std::string & text, bool bos);
-
-// load the tokens from encoder.json
-bool gpt_vocab_init(const std::string & fname, gpt_vocab & vocab);
+std::vector<llama_vocab::id> llama_tokenize(const llama_vocab & vocab, const std::string & text, bool bos);

 // sample next token given probabilities for each embedding
 //
 //   - consider only the top K tokens
 //   - from them, consider only the top tokens with cumulative probability > P
 //
-gpt_vocab::id llama_sample_top_p_top_k(
-        const gpt_vocab & vocab,
+llama_vocab::id llama_sample_top_p_top_k(
+        const llama_vocab & vocab,
        const float * logits,
-        std::vector<gpt_vocab::id> & last_n_tokens,
+        std::vector<llama_vocab::id> & last_n_tokens,
        double repeat_penalty,
        int top_k,
        double top_p,
@@ -96,7 +102,7 @@ gpt_vocab::id llama_sample_top_p_top_k(
        std::mt19937 & rng);

 // filer to top K tokens from list of logits
-void sample_top_k(std::vector<std::pair<double, gpt_vocab::id>> & logits_id, int top_k);
+void sample_top_k(std::vector<std::pair<double, llama_vocab::id>> & logits_id, int top_k);

 //
 // Quantization
Author	SHA1	Message	Date
anzz1	e6c9e0986c	Fix bin dir for win ci	2023-03-22 00:01:08 +02:00
Erik Scholz	01a297b099	specify build type for ctest on windows (#371 )	2023-03-21 23:34:25 +02:00
Georgi Gerganov	3366853e41	Add notice about pending change	2023-03-21 22:57:35 +02:00
Mathieu Nayrolles	3f9c6135e4	fix typo in chatLLaMa (#368 ) The prompt contains a typo where 'alound' is used instead of 'aloud'.	2023-03-21 22:52:27 +02:00
Georgi Gerganov	0f61352708	Update issue templates	2023-03-21 19:47:27 +02:00
Fabio R. Sluzala	353ec251a4	We could use std::unordered_map over std::map (#305 ) * Improve performance by changing std::map to std::unordered_map and std::map<id, token> id_to_token; to std::vector<token> id_to_token; * fix last commit on gpt_vocab_init add vocab.id_to_token.resize(vocab.token_to_id.size()); * Removed include <map> * Nest struct token score inside gpt_vocab * renamed token to tok	2023-03-21 19:21:50 +02:00
Matvey Soloviev	89d5d90f3b	Fix color codes emitting mid-UTF8 code. (#312 )	2023-03-21 19:11:01 +02:00
comex	16ffc013c6	Importer for GPTQ quantized LLaMA models (#301 ) * [WIP, broken] Importer for GPTQ quantized LLaMA models Based on: https://github.com/qwopqwop200/GPTQ-for-LLaMa Current status: Something is busted. The output starts out decent, but quickly degrades into gibberish. This doesn't happen with either the original GPTQ-for-LLaMa using the same weights, or llama.cpp when using weights quantized by its own quantizer. Is there a bug in the conversion script that somehow only comes into play with a large context size? I did notice one potential issue. It's clearly not the main cause of the gibberish, since it doesn't happen when using q4_1 weights quantized by llama.cpp itself, but it seems concerning. When doing a matrix multiplication of f16 * f32 => f32 or q4_1 * f32 => f32, at least when the multiplication is not done with BLAS, the intermediate results are stored in the smaller format rather than f32. This seems like an unnecessary waste of precision, especially in the q4_1 case. I was originally hoping to validate the results by matching the Python implementation's output exactly, but precision and non-associativity issues make this very difficult, including when performing matrix multiplications and, especially, computing norms. Anyway, design details: The models being imported store per-layer weights in essentially q4_1 format, although the addend and scale are shared across an entire row rather than every group of 32 weights. This script duplicates the addend and scale to match ggml's expectations, at the cost of wasting some memory. However, there are two differences which I accommodated changing the output format (and adding corresponding support to main.cpp) rather than having the script match the existing one: - The tok_embeddings and output weights (i.e. the weights that aren't per-layer) are f16 instead of q4_1. They could be converted to q4_1, and the impact of the loss of precision would probably be low, but this would rule out exactly matching the Python implementation's output for validation. - There is no sharding, since the input doesn't have it, and for a CPU-only implementation it seems more useful to avoid having to deal with multiple files. The new format is differentiated from existing q4_1 format by changing the 'f16' header flag to a new value, 4. That said, I think a cleaner approach would be to change main.cpp to support loading each tensor with an arbitrary sharding configuration and type rather than hardcoding specific combinations of types. So far I've wasted too much time debugging to try implementing this... * Add missing permutation. Now it works. --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 18:42:25 +02:00
Gary Linscott	486ae645fd	Compute perplexity over prompt (#270 ) * Compute perplexity over prompt * More accurate perplexity calculation - over all logits in the context window (so 512x more tokens!) * Output all perplexitiies * Add timing/ETA	2023-03-21 18:27:42 +02:00
Jean-Christophe Hoelt	3ab3e6582f	Add chatLLaMa script (#198 ) * Add chatLLaMa script * Fix shellcheck errors and do some cleanup * Move chatLLaMa script to `examples` directory * Reduce chatLLaMa context size to 2048 Ref `d7def1a752` * Include n_predict to 2048 in examples/chatLLaMa	2023-03-21 18:23:15 +02:00
Alex von Gluck IV	f157088cb7	makefile: Fix CPU feature detection on Haiku (#218 )	2023-03-21 18:21:06 +02:00
anzz1	c86ba036e6	Enable ANSI colors on Windows 10+ (#311 ) * Enable ANSI colors on Windows 10+ On older versions function will silently fail without any ill effects * Do not call SetConsoleMode if the mode is already set * Update main.cpp --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 18:14:46 +02:00
Georgi Gerganov	1daf4dd712	Minor style changes	2023-03-21 18:10:32 +02:00
Georgi Gerganov	dc6a845b85	Add chat.sh script	2023-03-21 18:09:46 +02:00
tjohnman	6a612959e1	Check for reverse prompt by characters instead of tokens (#292 ) (#330 ) * Check for reverse prompt by characters instead of tokens (#292) * Update main.cpp Wording. * Cleanup. * Remove unnecessary use of std::stringstream. --------- Co-authored-by: Johnman <tjohnman@github> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 18:05:06 +02:00
tjohnman	d5f56a5e5a	Check for reverse prompt by characters instead of tokens (#292 ) (#330 ) * Check for reverse prompt by characters instead of tokens (#292) * Update main.cpp Wording. * Cleanup. * Remove unnecessary use of std::stringstream. --------- Co-authored-by: Johnman <tjohnman@github> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 18:04:43 +02:00
Georgi Gerganov	3bfa3b43b7	Fix convert script, warnings alpaca instructions, default params	2023-03-21 17:59:16 +02:00
Kevin Lo	715d292ee0	Add OpenBSD support (#314 )	2023-03-21 17:50:09 +02:00
Mack Straight	c98ae02668	fix typo in comment (#318 )	2023-03-21 17:49:43 +02:00
Qingyou Meng	c3b2306b18	Makefile: slightly cleanup for Mac Intel; echo instead of run ./main -h (#335 )	2023-03-21 17:44:11 +02:00
anzz1	975d2cebf9	cmdline option for custom amount of model parts (--n_parts N) (#348 ) * cmdline option for custom amount of model parts (--n_parts N) * Update main.cpp --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 17:42:43 +02:00
Kevin Kwok	e0ffc861fa	Update IPFS links to quantized alpaca with new tokenizer format (#352 )	2023-03-21 17:34:49 +02:00
Georgi Gerganov	8f644a0a85	Change default repeat_penalty to 1.0 I feel this penalty is not really helping. Especially for the example from the README it makes results pretty bad	2023-03-21 17:32:14 +02:00
Georgi Gerganov	eb34620aec	Add tokenizer test + revert to C++11 (#355 ) * Add test-tokenizer-0 to do a few tokenizations - feel free to expand * Added option to convert-pth-to-ggml.py script to dump just the vocabulary * Added ./models/ggml-vocab.bin containing just LLaMA vocab data (used for tests) * Added utility to load vocabulary file from previous point (temporary implementation) * Avoid using std::string_view and drop back to C++11 (hope I didn't break something) * Rename gpt_vocab -> llama_vocab * All CMake binaries go into ./bin/ now	2023-03-21 17:29:41 +02:00
Casey Primozic	2e664f1ff4	Add initial AVX512 support for dot product on Linux (#320 ) * Update Makefile to detect AVX512 support and add compiler flags if it's available * Based on existing AVX2 implementation, dot product on one 32-value block of 4-bit quantized ints at a time * Perform 8 bit -> 16 bit sign extension and multiply+add on 32 values at time instead of 16 * Use built-in AVX512 horizontal reduce add to get sum at the end * Manual unrolling on inner dot product loop to reduce loop counter overhead	2023-03-21 15:35:42 +01:00
nusu-github	8cf9f34edd	Adding missing features of CMakeLists.txt & Refactoring (#131 ) * Functionality addition CMakeLists.txt Refactoring: 1. Simplify more options that are negation of negation. LLAMA_NO_ACCELERATE -> LLAMA_ACCELERATE 2. Changed to an optional expression instead of forcing to enable AVX2 in MSVC. 3. Make CMAKE_CXX_STANDARD, which is different from Makefile, the same. 4. Use add_compile_options instead of adding options to CMAKE_C_FLAGS. 5. Make utils use target_link_libraries instead of directly referencing code. Added features: 1. Added some options. LLAMA_STATIC_LINK,LLAMA_NATIVE,LLAMA_LTO,LLAMA_GPROF,LLAMA_OPENBLAS * Fix Accelerate link in CMake * Windows build Fix * C++11 to C++17 * Reflects C/C++ standard individually * Change the version to 3.12 --------- Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-21 01:37:16 +01:00
Ben Siraphob	bd4b46d6ba	Nix flake: set meta.mainProgram to llama	2023-03-20 22:50:22 +01:00
Qingyou Meng	6b6d5b5024	Fixed tokenizer.model not found error when model dir is symlink (#325 )	2023-03-20 19:33:10 +00:00
Mack Straight	a791a68b61	move file magic/version to header, print expected version (#319 )	2023-03-20 19:26:01 +00:00
Bernat Vadell	0f1b21cb90	Docker - Fix publish docker image in GitHub Registry (#235 ) * fix publish permission * try to fix docker pipeline using as password github_token & username repository_owner	2023-03-20 18:05:20 +01:00
Mack Straight	074bea2eb1	sentencepiece bpe compatible tokenizer (#252 ) * potential out of bounds read * fix quantize * style * Update convert-pth-to-ggml.py * mild cleanup * don't need the space-prefixing here rn since main.cpp already does it * new file magic + version header field * readme notice * missing newlines Co-authored-by: slaren <2141330+slaren@users.noreply.github.com>	2023-03-20 03:17:23 -07:00
Stephan Walter	5cb63e2493	Add tqdm to Python requirements (#293 ) * Add tqdm to Python requirements * Remove torchvision torchaudio, add requests	2023-03-20 09:24:11 +01:00
cocktailpeanut	da5303c1ea	bugfix: default should not be interactive (#304 )	2023-03-19 23:44:20 +02:00
Georgi Gerganov	4545539d71	Rename script	2023-03-19 21:58:51 +02:00
Georgi Gerganov	edeba28366	Add temporary helper script for Alpaca chat	2023-03-19 21:57:48 +02:00
Rickey Bowers Jr	5c19c70ba6	fix coloring of last `n_batch` of prompt, and refactor line input (#221 ) * fix coloring of last `n_batch` of prompt, and refactor line input * forgot the newline that needs to be sent to the model * (per #283) try to force flush of color reset in SIGINT handler	2023-03-19 19:44:30 +00:00
tjohnman	24568371ae	Support for multiple reverse prompts. (#299 ) Co-authored-by: Johnman <> Co-authored-by: Johnman <tjohnman@github>	2023-03-19 21:33:06 +02:00
Suaj Carrot	7392f1cd2c	Improved quantize script (#222 ) * Improved quantize script I improved the quantize script by adding error handling and allowing to select many models for quantization at once in the command line. I also converted it to Python for generalization as well as extensibility. * Fixes and improvements based on Matt's observations Fixed and improved many things in the script based on the reviews made by @mattsta. The parallelization suggestion is still to be revised, but code for it was still added (commented). * Small fixes to the previous commit * Corrected to use the original glob pattern The original Bash script uses a glob pattern to match files that have endings such as ...bin.0, ...bin.1, etc. That has been translated correctly to Python now. * Added support for Windows and updated README to use this script New code to set the name of the quantize script binary depending on the platform has been added (quantize.exe if working on Windows) and the README.md file has been updated to use this script instead of the Bash one. * Fixed a typo and removed shell=True in the subprocess.run call Fixed a typo regarding the new filenames of the quantized models and removed the shell=True parameter in the subprocess.run call as it was conflicting with the list of parameters. * Corrected previous commit * Small tweak: changed the name of the program in argparse This was making the automatic help message to be suggesting the program's usage as being literally "$ Quantization Script [arguments]". It should now be something like "$ python3 quantize.py [arguments]".	2023-03-19 20:38:44 +02:00
tjohnman	ad5fd5b60c	Make prompt randomization optional. (#300 ) Co-authored-by: Johnman <>	2023-03-19 20:36:19 +02:00
tjohnman	368d0c8a9e	Respect the maximum number of tokens in interactive. (#298 ) Co-authored-by: Johnman <johnman@github> Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-19 20:31:17 +02:00
slaren	50fae10d03	Add --ignore-eos parameter (#181 ) Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>	2023-03-19 20:22:48 +02:00