gguf : new file format with flexible meta data (beta) (#2398)

* gguf : first API pass

* gguf : read header + meta data

* gguf : read tensor info

* gguf : initial model loading - not tested

* gguf : add gguf_get_tensor_name()

* gguf : do not support passing existing ggml_context to gguf_init

* gguf : simplify gguf_get_val

* gguf : gguf.c is now part of ggml.c

* gguf : read / write sample models

* gguf : add comments

* refactor : reduce code duplication and better API (#2415)

* gguf : expose the gguf_type enum through the API for now

* gguf : add array support

* gguf.py : some code style changes

* convert.py : start a new simplified implementation by removing old stuff

* convert.py : remove GGML vocab + other obsolete stuff

* GGUF : write tensor (#2426)

* WIP: Write tensor

* GGUF : Support writing tensors in Python

* refactor : rm unused import and upd todos

* fix : fix errors upd writing example

* rm example.gguf

* gitignore *.gguf

* undo formatting

* gguf : add gguf_find_key (#2438)

* gguf.cpp : find key example

* ggml.h : add gguf_find_key

* ggml.c : add gguf_find_key

* gguf : fix writing tensors

* gguf : do not hardcode tensor names to read

* gguf : write sample tensors to read

* gguf : add tokenization constants

* quick and dirty conversion example

* gguf : fix writing gguf arrays

* gguf : write tensors one by one and code reuse

* gguf : fix writing gguf arrays

* gguf : write tensors one by one

* gguf : write tensors one by one

* gguf : write tokenizer data

* gguf : upd gguf conversion script

* Update convert-llama-h5-to-gguf.py

* gguf : handle already encoded string

* ggml.h : get array str and f32

* ggml.c : get arr str and f32

* gguf.py : support any type

* Update convert-llama-h5-to-gguf.py

* gguf : fix set is not subscriptable

* gguf : update convert-llama-h5-to-gguf.py

* constants.py : add layer norm eps

* gguf.py : add layer norm eps and merges

* ggml.h : increase GGML_MAX_NAME to 64

* ggml.c : add gguf_get_arr_n

* Update convert-llama-h5-to-gguf.py

* add gptneox gguf example

* Makefile : add gptneox gguf example

* Update convert-llama-h5-to-gguf.py

* add gptneox gguf example

* Update convert-llama-h5-to-gguf.py

* Update convert-gptneox-h5-to-gguf.py

* Update convert-gptneox-h5-to-gguf.py

* Update convert-llama-h5-to-gguf.py

* gguf : support custom alignment value

* gguf : fix typo in function call

* gguf : mmap tensor data example

* fix : update convert-llama-h5-to-gguf.py

* Update convert-llama-h5-to-gguf.py

* convert-gptneox-h5-to-gguf.py : Special tokens

* gptneox-main.cpp : special tokens

* Update gptneox-main.cpp

* constants.py : special tokens

* gguf.py : accumulate kv and tensor info data + special tokens

* convert-gptneox-h5-to-gguf.py : accumulate kv and ti + special tokens

* gguf : gguf counterpart of llama-util.h

* gguf-util.h : update note

* convert-llama-h5-to-gguf.py : accumulate kv / ti + special tokens

* convert-llama-h5-to-gguf.py : special tokens

* Delete gptneox-common.cpp

* Delete gptneox-common.h

* convert-gptneox-h5-to-gguf.py : gpt2bpe tokenizer

* gptneox-main.cpp : gpt2 bpe tokenizer

* gpt2 bpe tokenizer (handles merges and unicode)

* Makefile : remove gptneox-common

* gguf.py : bytesarray for gpt2bpe tokenizer

* cmpnct_gpt2bpe.hpp : comments

* gguf.py : use custom alignment if present

* gguf : minor stuff

* Update gptneox-main.cpp

* map tensor names

* convert-gptneox-h5-to-gguf.py : map tensor names

* convert-llama-h5-to-gguf.py : map tensor names

* gptneox-main.cpp : map tensor names

* gguf : start implementing libllama in GGUF (WIP)

* gguf : start implementing libllama in GGUF (WIP)

* rm binary commited by mistake

* upd .gitignore

* gguf : calculate n_mult

* gguf :  inference with 7B model working (WIP)

* gguf : rm deprecated function

* gguf : start implementing gguf_file_saver (WIP)

* gguf : start implementing gguf_file_saver (WIP)

* gguf : start implementing gguf_file_saver (WIP)

* gguf : add gguf_get_kv_type

* gguf : add gguf_get_kv_type

* gguf : write metadata in gguf_file_saver (WIP)

* gguf : write metadata in gguf_file_saver (WIP)

* gguf : write metadata in gguf_file_saver

* gguf : rm references to old file formats

* gguf : shorter name for member variable

* gguf : rm redundant method

* gguf : get rid of n_mult, read n_ff from file

* Update gguf_tensor_map.py

* Update gptneox-main.cpp

* gguf : rm references to old file magics

* gguf : start implementing quantization (WIP)

* gguf : start implementing quantization (WIP)

* gguf : start implementing quantization (WIP)

* gguf : start implementing quantization (WIP)

* gguf : start implementing quantization (WIP)

* gguf : start implementing quantization (WIP)

* gguf : quantization is working

* gguf : roper closing of file

* gguf.py : no need to convert tensors twice

* convert-gptneox-h5-to-gguf.py : no need to convert tensors twice

* convert-llama-h5-to-gguf.py : no need to convert tensors twice

* convert-gptneox-h5-to-gguf.py : simplify nbytes

* convert-llama-h5-to-gguf.py : simplify nbytes

* gptneox-main.cpp : n_layer --> n_block

* constants.py : n_layer --> n_block

* gguf.py : n_layer --> n_block

* convert-gptneox-h5-to-gguf.py : n_layer --> n_block

* convert-llama-h5-to-gguf.py : n_layer --> n_block

* gptneox-main.cpp : n_layer --> n_block

* Update gguf_tensor_map.py

* convert-gptneox-h5-to-gguf.py : load model in parts to save memory

* convert-llama-h5-to-gguf.py : load model in parts to save memory

* convert : write more metadata for LLaMA

* convert : rm quantization version

* convert-gptneox-h5-to-gguf.py : add file_type key

* gptneox-main.cpp : add file_type key

* fix conflicts

* gguf : add todos and comments

* convert-gptneox-h5-to-gguf.py : tensor name map changes

* Create gguf_namemap.py : tensor name map changes

* Delete gguf_tensor_map.py

* gptneox-main.cpp : tensor name map changes

* convert-llama-h5-to-gguf.py : fixes

* gguf.py : dont add empty strings

* simple : minor style changes

* gguf : use UNIX line ending

* Create convert-llama-7b-pth-to-gguf.py

* llama : sync gguf-llama.cpp with latest llama.cpp (#2608)

* llama : sync gguf-llama.cpp with latest llama.cpp

* minor : indentation + assert

* llama : refactor gguf_buffer and gguf_ctx_buffer

* llama : minor

* gitignore : add gptneox-main

* llama : tokenizer fixes (#2549)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* convert : update convert-new.py with tokenizer fixes (#2614)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)

* llama : sync gguf-llama with llama (#2613)

* llama : sync gguf-llama with llama

* tests : fix build + warnings (test-tokenizer-1 still fails)

* tests : fix wstring_convert

* convert : fix layer names

* llama : sync gguf-llama.cpp

* convert : update HF converter to new tokenizer voodoo magics

* llama : update tokenizer style

* convert-llama-h5-to-gguf.py : add token types

* constants.py : add token types

* gguf.py : add token types

* convert-llama-7b-pth-to-gguf.py : add token types

* gguf-llama.cpp :  fix n_head_kv

* convert-llama-h5-to-gguf.py : add 70b gqa support

* gguf.py : add tensor data layout

* convert-llama-h5-to-gguf.py : add tensor data layout

* convert-llama-7b-pth-to-gguf.py : add tensor data layout

* gptneox-main.cpp : add tensor data layout

* convert-llama-h5-to-gguf.py : clarify the reverse permute

* llama : refactor model loading code (#2620)

* llama : style formatting + remove helper methods

* llama : fix quantization using gguf tool

* llama : simplify gguf_file_saver

* llama : fix method names

* llama : simplify write_header()

* llama : no need to pass full file loader to the file saver

just gguf_ctx

* llama : gguf_file_saver write I32

* llama : refactor tensor names (#2622)

* gguf: update tensor names searched in quantization

* gguf : define tensor names as constants

* gguf : initial write API (not tested yet)

* gguf : write to file API (not tested)

* gguf : initial write API ready + example

* gguf : fix header write

* gguf : fixes + simplify example + add ggml_nbytes_pad()

* gguf : minor

* llama : replace gguf_file_saver with new gguf write API

* gguf : streaming support when writing files

* gguf : remove oboslete write methods

* gguf : remove obosolete gguf_get_arr_xxx API

* llama : simplify gguf_file_loader

* llama : move hparams and vocab from gguf_file_loader to llama_model_loader

* llama : merge gguf-util.h in llama.cpp

* llama : reorder definitions in .cpp to match .h

* llama : minor simplifications

* llama : refactor llama_model_loader (WIP)

wip : remove ggml_ctx from llama_model_loader

wip : merge gguf_file_loader in llama_model_loader

* llama : fix shape prints

* llama : fix Windows build + fix norm_rms_eps key

* llama : throw error on missing KV paris in model meta data

* llama : improve printing + log meta data

* llama : switch print order of meta data

---------

Co-authored-by: M. Yusuf Sarıgöz <yusufsarigoz@gmail.com>

* gguf : deduplicate (#2629)

* gguf : better type names

* dedup : CPU + Metal is working

* ggml : fix warnings about unused results

* llama.cpp : fix line feed and compiler warning

* llama : fix strncpy warning + note token_to_str does not write null

* llama : restore the original load/save session implementation

Will migrate this to GGUF in the future

* convert-llama-h5-to-gguf.py : support alt ctx param name

* ggml : assert when using ggml_mul with non-F32 src1

* examples : dedup simple

---------

Co-authored-by: klosax <131523366+klosax@users.noreply.github.com>

* gguf.py : merge all files in gguf.py

* convert-new.py : pick #2427 for HF 70B support

* examples/gguf : no need to keep q option for quantization any more

* llama.cpp : print actual model size

* llama.cpp : use ggml_elements()

* convert-new.py : output gguf (#2635)

* convert-new.py : output gguf (WIP)

* convert-new.py : add gguf key-value pairs

* llama : add hparams.ctx_train + no longer print ftype

* convert-new.py : minor fixes

* convert-new.py : vocab-only option should work now

* llama : fix tokenizer to use llama_char_to_byte

* tests : add new ggml-vocab-llama.gguf

* convert-new.py : tensor name mapping

* convert-new.py : add map for skipping tensor serialization

* convert-new.py : convert script now works

* gguf.py : pick some of the refactoring from #2644

* convert-new.py : minor fixes

* convert.py : update to support GGUF output

* Revert "ci : disable CI temporary to not waste energy"

This reverts commit 7e82d25f40386540c2c15226300ad998ecd871ea.

* convert.py : n_head_kv optional and .gguf file extension

* convert.py : better always have n_head_kv and default it to n_head

* llama : sync with recent PRs on master

* editorconfig : ignore models folder

ggml-ci

* ci : update ".bin" to ".gguf" extension

ggml-ci

* llama : fix llama_model_loader memory leak

* gptneox : move as a WIP example

* llama : fix lambda capture

ggml-ci

* ggml : fix bug in gguf_set_kv

ggml-ci

* common.h : .bin --> .gguf

* quantize-stats.cpp : .bin --> .gguf

* convert.py : fix HF tensor permuting / unpacking

ggml-ci

* llama.cpp : typo

* llama : throw error if gguf fails to init from file

ggml-ci

* llama : fix tensor name grepping during quantization

ggml-ci

* gguf.py : write tensors in a single pass (#2644)

* gguf : single pass for writing tensors + refactoring writer

* gguf : single pass for writing tensors + refactoring writer

* gguf : single pass for writing tensors + refactoring writer

* gguf : style fixes in simple conversion script

* gguf : refactor gptneox conversion script

* gguf : rename h5 to hf (for HuggingFace)

* gguf : refactor pth to gguf conversion script

* gguf : rm file_type key and method

* gguf.py : fix vertical alignment

* gguf.py : indentation

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>

* convert-gptneox-hf-to-gguf.py : fixes

* gguf.py : gptneox mapping

* convert-llama-hf-to-gguf.py : fixes

* convert-llama-7b-pth-to-gguf.py : fixes

* ggml.h : reverse GGUF_MAGIC

* gguf.py : reverse GGUF_MAGIC

* test-tokenizer-0.cpp : fix warning

* llama.cpp : print kv general.name

* llama.cpp : get special token kv and linefeed token id

* llama : print number of tensors per type + print arch + style

* tests : update vocab file with new magic

* editorconfig : fix whitespaces

* llama : re-order functions

* llama : remove C++ API + reorganize common source in /common dir

* llama : minor API updates

* llama : avoid hardcoded special tokens

* llama : fix MPI build

ggml-ci

* llama : introduce enum llama_vocab_type + remove hardcoded string constants

* convert-falcon-hf-to-gguf.py : falcon HF --> gguf conversion, not tested

* falcon-main.cpp : falcon inference example

* convert-falcon-hf-to-gguf.py : remove extra kv

* convert-gptneox-hf-to-gguf.py : remove extra kv

* convert-llama-7b-pth-to-gguf.py : remove extra kv

* convert-llama-hf-to-gguf.py : remove extra kv

* gguf.py : fix for falcon 40b

* falcon-main.cpp : fix for falcon 40b

* convert-falcon-hf-to-gguf.py : update ref

* convert-falcon-hf-to-gguf.py : add tensor data layout

* cmpnct_gpt2bpe.hpp : fixes

* falcon-main.cpp : fixes

* gptneox-main.cpp : fixes

* cmpnct_gpt2bpe.hpp : remove non-general stuff

* Update examples/server/README.md

Co-authored-by: slaren <slarengh@gmail.com>

* cmpnct_gpt2bpe.hpp : cleanup

* convert-llama-hf-to-gguf.py : special tokens

* convert-llama-7b-pth-to-gguf.py : special tokens

* convert-permute-debug.py : permute debug print

* convert-permute-debug-master.py : permute debug for master

* convert-permute-debug.py : change permute type of attn_q

* convert.py : 70b model working (change attn_q permute)

* Delete convert-permute-debug-master.py

* Delete convert-permute-debug.py

* convert-llama-hf-to-gguf.py : fix attn_q permute

* gguf.py : fix rope scale kv

* convert-llama-hf-to-gguf.py : rope scale and added tokens

* convert-llama-7b-pth-to-gguf.py : rope scale and added tokens

* llama.cpp : use rope scale kv

* convert-llama-7b-pth-to-gguf.py : rope scale fix

* convert-llama-hf-to-gguf.py : rope scale fix

* py : fix whitespace

* gguf : add Python script to convert GGMLv3 LLaMA models to GGUF (#2682)

* First pass at converting GGMLv3 LLaMA models to GGUF

* Cleanups, better output during conversion

* Fix vocab space conversion logic

* More vocab conversion fixes

* Add description to converted GGUF files

* Improve help text, expand warning

* Allow specifying name and description for output GGUF

* Allow overriding vocab and hyperparams from original model metadata

* Use correct params override var name

* Fix wrong type size for Q8_K

Better handling of original style metadata

* Set default value for gguf add_tensor raw_shape KW arg

* llama : improve token type support (#2668)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)

* Improved tokenizer test

But does it work on MacOS?

* Improve token type support

- Added @klosax code to convert.py
- Improved token type support in vocabulary

* Exclude platform dependent tests

* More sentencepiece compatibility by eliminating magic numbers

* Restored accidentally removed comment

* llama : add API for token type

ggml-ci

* tests : use new tokenizer type API (#2692)

* Merge tokenizer fixes into the gguf branch.

* Add test vocabularies

* Adapt convert-new.py (and fix a clang-cl compiler error on windows)

* Improved tokenizer test

But does it work on MacOS?

* Improve token type support

- Added @klosax code to convert.py
- Improved token type support in vocabulary

* Exclude platform dependent tests

* More sentencepiece compatibility by eliminating magic numbers

* Restored accidentally removed comment

* Improve commentary

* Use token type API in test-tokenizer-1.cpp

* py : cosmetics

* readme : add notice about new file format

ggml-ci

---------

Co-authored-by: M. Yusuf Sarıgöz <yusufsarigoz@gmail.com>
Co-authored-by: klosax <131523366+klosax@users.noreply.github.com>
Co-authored-by: goerch <jhr.walter@t-online.de>
Co-authored-by: slaren <slarengh@gmail.com>
Co-authored-by: Kerfuffle <44031344+KerfuffleV2@users.noreply.github.com>

1 files changed, 1009 insertions, 4 deletions

diff --git a/ggml.c b/ggml.c
index 44c43b42..c917d73c 100644
--- a/ggml.c
+++ b/ggml.c
@@ -213,10 +213,10 @@ inline static void * ggml_aligned_malloc(size_t size) {
                 error_desc = "insufficient memory";
                 break;
         }
-        GGML_PRINT("%s: %s (attempted to allocate %6.2f MB)\n",
-            __func__, error_desc, size/(1024.0*1024.0));
+        GGML_PRINT("%s: %s (attempted to allocate %6.2f MB)\n", __func__, error_desc, size/(1024.0*1024.0));
         return NULL;
     }
+
     return aligned_memory;
 }
 #define GGML_ALIGNED_MALLOC(size)  ggml_aligned_malloc(size)
@@ -4091,7 +4091,11 @@ size_t ggml_nbytes(const struct ggml_tensor * tensor) {
     //
     // is enough, but just in case, adding the second part
 
-    return GGML_PAD(MAX(tensor->ne[3]*tensor->nb[3], ggml_nelements(tensor)*ggml_type_size(tensor->type))/ggml_blck_size(tensor->type), GGML_MEM_ALIGN);
+    return MAX(tensor->ne[3]*tensor->nb[3], (ggml_nelements(tensor)*ggml_type_size(tensor->type))/ggml_blck_size(tensor->type));
+}
+
+size_t ggml_nbytes_pad(const struct ggml_tensor * tensor) {
+    return GGML_PAD(ggml_nbytes(tensor), GGML_MEM_ALIGN);
 }
 
 size_t ggml_nbytes_split(const struct ggml_tensor * tensor, int nrows_split) {
@@ -9118,6 +9122,8 @@ static void ggml_compute_forward_mul(
         const struct ggml_tensor * src0,
         const struct ggml_tensor * src1,
         struct ggml_tensor * dst) {
+    GGML_ASSERT(src1->type == GGML_TYPE_F32 && "only f32 src1 supported for now");
+
     switch (src0->type) {
         case GGML_TYPE_F32:
             {
@@ -16881,7 +16887,7 @@ void ggml_graph_export(const struct ggml_cgraph * cgraph, const char * fname) {
     // compute size of intermediate results
     // TODO: does not take into account scratch buffers !!!!
     for (int i = 0; i < cgraph->n_nodes; ++i) {
-        size_eval += ggml_nbytes(cgraph->nodes[i]);
+        size_eval += ggml_nbytes_pad(cgraph->nodes[i]);
     }
 
     // print
@@ -18542,6 +18548,1005 @@ size_t ggml_quantize_chunk(enum ggml_type type, const float * src, void * dst, i
 
 ////////////////////////////////////////////////////////////////////////////////
 
+struct gguf_str {
+    uint32_t n;
+    char * data;
+};
+
+static const size_t GGUF_TYPE_SIZE[GGUF_TYPE_COUNT] = {
+    [GGUF_TYPE_UINT8]   = sizeof(uint8_t),
+    [GGUF_TYPE_INT8]    = sizeof(int8_t),
+    [GGUF_TYPE_UINT16]  = sizeof(uint16_t),
+    [GGUF_TYPE_INT16]   = sizeof(int16_t),
+    [GGUF_TYPE_UINT32]  = sizeof(uint32_t),
+    [GGUF_TYPE_INT32]   = sizeof(int32_t),
+    [GGUF_TYPE_FLOAT32] = sizeof(float),
+    [GGUF_TYPE_BOOL]    = sizeof(bool),
+    [GGUF_TYPE_STRING]  = sizeof(struct gguf_str),
+    [GGUF_TYPE_ARRAY]   = 0, // undefined
+};
+static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
+
+static const char * GGUF_TYPE_NAME[GGUF_TYPE_COUNT] = {
+    [GGUF_TYPE_UINT8]   = "u8",
+    [GGUF_TYPE_INT8]    = "i8",
+    [GGUF_TYPE_UINT16]  = "u16",
+    [GGUF_TYPE_INT16]   = "i16",
+    [GGUF_TYPE_UINT32]  = "u32",
+    [GGUF_TYPE_INT32]   = "i32",
+    [GGUF_TYPE_FLOAT32] = "f32",
+    [GGUF_TYPE_BOOL]    = "bool",
+    [GGUF_TYPE_STRING]  = "str",
+    [GGUF_TYPE_ARRAY]   = "arr",
+};
+static_assert(GGUF_TYPE_COUNT == 10, "GGUF_TYPE_COUNT != 10");
+
+union gguf_value {
+    uint8_t  uint8;
+    int8_t   int8;
+    uint16_t uint16;
+    int16_t  int16;
+    uint32_t uint32;
+    int32_t  int32;
+    float    float32;
+    bool     bool_;
+
+    struct gguf_str str;
+
+    struct {
+        enum gguf_type type;
+
+        uint32_t n;
+        void * data;
+    } arr;
+};
+
+struct gguf_kv {
+    struct gguf_str key;
+
+    uint32_t n_bytes; // TODO: is this actually needed?
+
+    enum  gguf_type  type;
+    union gguf_value value;
+};
+
+struct gguf_header {
+    uint32_t magic;
+    uint32_t version;
+    uint32_t n_tensors;
+    uint32_t n_kv;
+};
+
+struct gguf_tensor_info {
+    struct gguf_str name;
+
+    uint32_t n_dims;
+    uint32_t ne[GGML_MAX_DIMS];
+
+    enum ggml_type type;
+
+    uint64_t offset; // offset from start of `data`, must be a multiple of `ALIGNMENT`
+
+    // for writing API
+    const void * data;
+    size_t size;
+};
+
+struct gguf_context {
+    struct gguf_header header;
+
+    struct gguf_kv          * kv;
+    struct gguf_tensor_info * infos;
+
+    size_t alignment;
+    size_t offset;    // offset of `data` from beginning of file
+    size_t size;      // size of `data` in bytes
+
+    //uint8_t * padding;
+    void * data;
+};
+
+static bool gguf_fread_el(FILE * file, void * dst, size_t size, size_t * offset) {
+    const size_t n = fread(dst, 1, size, file);
+    *offset += n;
+    return n == size;
+}
+
+static bool gguf_fread_str(FILE * file, struct gguf_str * p, size_t * offset) {
+    p->n    = 0;
+    p->data = NULL;
+
+    bool ok = true;
+
+    // TODO: how to avoid mallocs for strings?
+    ok = ok && gguf_fread_el(file, &p->n,    sizeof(p->n), offset); p->data = calloc(p->n + 1, 1);
+    ok = ok && gguf_fread_el(file,  p->data, p->n,         offset);
+
+    return ok;
+}
+
+struct gguf_context * gguf_init_empty(void) {
+    struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
+
+    ctx->header.magic     = GGUF_MAGIC;
+    ctx->header.version   = GGUF_VERSION;
+    ctx->header.n_tensors = 0;
+    ctx->header.n_kv      = 0;
+
+    ctx->kv    = NULL;
+    ctx->infos = NULL;
+
+    ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
+    ctx->offset    = 0;
+    ctx->size      = 0;
+
+    ctx->data = NULL;
+
+    return ctx;
+}
+
+struct gguf_context * gguf_init_from_file(const char * fname, struct gguf_init_params params) {
+    FILE * file = fopen(fname, "rb");
+    if (!file) {
+        return NULL;
+    }
+
+    // offset from start of file
+    size_t offset = 0;
+
+    uint32_t magic = 0;
+
+    // check the magic before making allocations
+    {
+        gguf_fread_el(file, &magic, sizeof(magic), &offset);
+
+        if (magic != GGUF_MAGIC) {
+            fprintf(stderr, "%s: invalid magic number %08x\n", __func__, magic);
+            fclose(file);
+            return NULL;
+        }
+    }
+
+    bool ok = true;
+
+    struct gguf_context * ctx = GGML_ALIGNED_MALLOC(sizeof(struct gguf_context));
+
+    // read the header
+    {
+        ctx->header.magic = magic;
+
+        ctx->kv    = NULL;
+        ctx->infos = NULL;
+        ctx->data  = NULL;
+
+        ok = ok && gguf_fread_el(file, &ctx->header.version,   sizeof(ctx->header.version),   &offset);
+        ok = ok && gguf_fread_el(file, &ctx->header.n_tensors, sizeof(ctx->header.n_tensors), &offset);
+        ok = ok && gguf_fread_el(file, &ctx->header.n_kv,      sizeof(ctx->header.n_kv),      &offset);
+
+        if (!ok) {
+            fprintf(stderr, "%s: failed to read header\n", __func__);
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
+        }
+    }
+
+    // read the kv pairs
+    {
+        ctx->kv = GGML_ALIGNED_MALLOC(ctx->header.n_kv * sizeof(struct gguf_kv));
+
+        for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+            struct gguf_kv * kv = &ctx->kv[i];
+
+            //fprintf(stderr, "%s: reading kv %d\n", __func__, i);
+
+            ok = ok && gguf_fread_str(file, &kv->key,                          &offset);
+          //ok = ok && gguf_fread_el (file, &kv->n_bytes, sizeof(kv->n_bytes), &offset);
+            ok = ok && gguf_fread_el (file, &kv->type,    sizeof(kv->type),    &offset);
+
+            //fprintf(stderr, "%s: reading kv with key %s\n", __func__, kv->key.data);
+
+            switch (kv->type) {
+                case GGUF_TYPE_UINT8:   ok = ok && gguf_fread_el (file, &kv->value.uint8,   sizeof(kv->value.uint8),   &offset); break;
+                case GGUF_TYPE_INT8:    ok = ok && gguf_fread_el (file, &kv->value.int8,    sizeof(kv->value.int8),    &offset); break;
+                case GGUF_TYPE_UINT16:  ok = ok && gguf_fread_el (file, &kv->value.uint16,  sizeof(kv->value.uint16),  &offset); break;
+                case GGUF_TYPE_INT16:   ok = ok && gguf_fread_el (file, &kv->value.int16,   sizeof(kv->value.int16),   &offset); break;
+                case GGUF_TYPE_UINT32:  ok = ok && gguf_fread_el (file, &kv->value.uint32,  sizeof(kv->value.uint32),  &offset); break;
+                case GGUF_TYPE_INT32:   ok = ok && gguf_fread_el (file, &kv->value.int32,   sizeof(kv->value.int32),   &offset); break;
+                case GGUF_TYPE_FLOAT32: ok = ok && gguf_fread_el (file, &kv->value.float32, sizeof(kv->value.float32), &offset); break;
+                case GGUF_TYPE_BOOL:    ok = ok && gguf_fread_el (file, &kv->value.bool_,   sizeof(kv->value.bool_),   &offset); break;
+                case GGUF_TYPE_STRING:  ok = ok && gguf_fread_str(file, &kv->value.str,                                &offset); break;
+                case GGUF_TYPE_ARRAY:
+                    {
+                        ok = ok && gguf_fread_el(file, &kv->value.arr.type, sizeof(kv->value.arr.type), &offset);
+                        ok = ok && gguf_fread_el(file, &kv->value.arr.n,    sizeof(kv->value.arr.n),    &offset);
+
+                        switch (kv->value.arr.type) {
+                            case GGUF_TYPE_UINT8:
+                            case GGUF_TYPE_INT8:
+                            case GGUF_TYPE_UINT16:
+                            case GGUF_TYPE_INT16:
+                            case GGUF_TYPE_UINT32:
+                            case GGUF_TYPE_INT32:
+                            case GGUF_TYPE_FLOAT32:
+                            case GGUF_TYPE_BOOL:
+                                {
+                                    kv->value.arr.data = malloc(kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type]);
+                                    ok = ok && gguf_fread_el(file, kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type], &offset);
+                                } break;
+                            case GGUF_TYPE_STRING:
+                                {
+                                    kv->value.arr.data = malloc(kv->value.arr.n * sizeof(struct gguf_str));
+                                    for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                                        ok = ok && gguf_fread_str(file, &((struct gguf_str *) kv->value.arr.data)[j], &offset);
+                                    }
+                                } break;
+                            case GGUF_TYPE_ARRAY:
+                            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type"); break;
+                        };
+                    } break;
+                case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
+            };
+
+            if (!ok) {
+                break;
+            }
+        }
+
+        if (!ok) {
+            fprintf(stderr, "%s: failed to read key-value pairs\n", __func__);
+            fclose(file);
+            gguf_free(ctx);
+            return NULL;
+        }
+    }
+
+    // read the tensor infos
+    {
+        ctx->infos = GGML_ALIGNED_MALLOC(ctx->header.n_tensors * sizeof(struct gguf_tensor_info));
+
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
+
+            for (int j = 0; j < GGML_MAX_DIMS; ++j) {
+                info->ne[j] = 1;
+            }
+
+            ok = ok && gguf_fread_str(file, &info->name,                          &offset);
+            ok = ok && gguf_fread_el (file, &info->n_dims, sizeof(info->n_dims),  &offset);
+            for (uint32_t j = 0; j < info->n_dims; ++j) {
+                ok = ok && gguf_fread_el(file, &info->ne[j], sizeof(info->ne[j]), &offset);
+            }
+            ok = ok && gguf_fread_el (file, &info->type,   sizeof(info->type),    &offset);
+            ok = ok && gguf_fread_el (file, &info->offset, sizeof(info->offset),  &offset);
+
+            if (!ok) {
+                fprintf(stderr, "%s: failed to read tensor info\n", __func__);
+                fclose(file);
+                gguf_free(ctx);
+                return NULL;
+            }
+        }
+    }
+
+    ctx->alignment = GGUF_DEFAULT_ALIGNMENT;
+
+    int alignment_idx = gguf_find_key(ctx, "general.alignment");
+    if (alignment_idx != -1) {
+        ctx->alignment = gguf_get_val_u32(ctx, alignment_idx);
+    }
+
+    // we require the data section to be aligned, so take into account any padding
+    {
+        const size_t offset_pad = offset % ctx->alignment;
+
+        if (offset_pad != 0) {
+            offset += ctx->alignment - offset_pad;
+            fseek(file, offset, SEEK_SET);
+        }
+    }
+
+    // store the current file offset - this is where the data section starts
+    ctx->offset = offset;
+
+    // compute the total size of the data section, taking into account the alignment
+    {
+        ctx->size = 0;
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
+
+            const int64_t ne =
+                (int64_t) info->ne[0] *
+                (int64_t) info->ne[1] *
+                (int64_t) info->ne[2] *
+                (int64_t) info->ne[3];
+
+            if (ne % ggml_blck_size(info->type) != 0) {
+                fprintf(stderr, "%s: tensor '%s' number of elements (%" PRId64 ") is not a multiple of block size (%d)\n",
+                        __func__, info->name.data, ne, ggml_blck_size(info->type));
+                fclose(file);
+                gguf_free(ctx);
+                return NULL;
+            }
+
+            const size_t size_cur = (ne*ggml_type_size(info->type))/ggml_blck_size(info->type);
+
+            ctx->size += GGML_PAD(size_cur, ctx->alignment);
+        }
+    }
+
+    // load the tensor data only if requested
+    if (params.ctx != NULL) {
+        // if the provided gguf_context is no_alloc, then we create "empty" tensors and do not read the binary blob
+        // otherwise, we load the binary blob into the created ggml_context as well, and point the "data" members of
+        // the ggml_tensor structs to the appropriate locations in the binary blob
+
+        // compute the exact size needed for the new ggml_context
+        const size_t mem_size =
+            params.no_alloc ?
+            (ctx->header.n_tensors    )*ggml_tensor_overhead() :
+            (ctx->header.n_tensors + 1)*ggml_tensor_overhead() + ctx->size;
+
+        struct ggml_init_params pdata = {
+            .mem_size   = mem_size,
+            .mem_buffer = NULL,
+            .no_alloc   = params.no_alloc,
+        };
+
+        *params.ctx = ggml_init(pdata);
+
+        struct ggml_context * ctx_data = *params.ctx;
+
+        struct ggml_tensor * data = NULL;
+
+        if (params.no_alloc == false) {
+            data = ggml_new_tensor_1d(ctx_data, GGML_TYPE_I8, ctx->size);
+
+            ok = ok && data != NULL;
+
+            // read the binary blob with the tensor data
+            ok = ok && gguf_fread_el(file, data->data, ctx->size, &offset);
+
+            if (!ok) {
+                fprintf(stderr, "%s: failed to read tensor data\n", __func__);
+                fclose(file);
+                ggml_free(ctx_data);
+                gguf_free(ctx);
+                return NULL;
+            }
+
+            ctx->data = data->data;
+        }
+
+        ggml_set_no_alloc(ctx_data, true);
+
+        // create the tensors
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            const int64_t ne[GGML_MAX_DIMS] = {
+                ctx->infos[i].ne[0],
+                ctx->infos[i].ne[1],
+                ctx->infos[i].ne[2],
+                ctx->infos[i].ne[3],
+            };
+
+            struct ggml_tensor * cur = ggml_new_tensor(ctx_data, ctx->infos[i].type, ctx->infos[i].n_dims, ne);
+
+            ok = ok && cur != NULL;
+
+            ggml_set_name(cur, ctx->infos[i].name.data);
+
+            if (!ok) {
+                break;
+            }
+
+            // point the data member to the appropriate location in the binary blob using the tensor infos
+            if (params.no_alloc == false) {
+              //cur->data = (char *) data->data + ctx->infos[i].offset - ctx->offset; // offset from start of file
+                cur->data = (char *) data->data + ctx->infos[i].offset;               // offset from data
+            }
+        }
+
+        if (!ok) {
+            fprintf(stderr, "%s: failed to read the tensor data\n", __func__);
+            fclose(file);
+            ggml_free(ctx_data);
+            gguf_free(ctx);
+            return NULL;
+        }
+
+        ggml_set_no_alloc(ctx_data, params.no_alloc);
+    }
+
+    fclose(file);
+
+    return ctx;
+}
+
+void gguf_free(struct gguf_context * ctx) {
+    if (ctx == NULL) {
+        return;
+    }
+
+    if (ctx->kv) {
+        // free string memory - not great..
+        for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+            struct gguf_kv * kv = &ctx->kv[i];
+
+            if (kv->key.data) {
+                free(kv->key.data);
+            }
+
+            if (kv->type == GGUF_TYPE_STRING) {
+                if (kv->value.str.data) {
+                    free(kv->value.str.data);
+                }
+            }
+
+            if (kv->type == GGUF_TYPE_ARRAY) {
+                if (kv->value.arr.data) {
+                    if (kv->value.arr.type == GGUF_TYPE_STRING) {
+                        for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                            struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[j];
+                            if (str->data) {
+                                free(str->data);
+                            }
+                        }
+                    }
+                    free(kv->value.arr.data);
+                }
+            }
+        }
+
+        GGML_ALIGNED_FREE(ctx->kv);
+    }
+
+    if (ctx->infos) {
+        for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+            struct gguf_tensor_info * info = &ctx->infos[i];
+
+            if (info->name.data) {
+                free(info->name.data);
+            }
+        }
+
+        GGML_ALIGNED_FREE(ctx->infos);
+    }
+
+    GGML_ALIGNED_FREE(ctx);
+}
+
+const char * gguf_type_name(enum gguf_type type) {
+    return GGUF_TYPE_NAME[type];
+}
+
+int gguf_get_version(struct gguf_context * ctx) {
+    return ctx->header.version;
+}
+
+size_t gguf_get_alignment(struct gguf_context * ctx) {
+    return ctx->alignment;
+}
+
+size_t gguf_get_data_offset(struct gguf_context * ctx) {
+    return ctx->offset;
+}
+
+void * gguf_get_data(struct gguf_context * ctx) {
+    return ctx->data;
+}
+
+int gguf_get_n_kv(struct gguf_context * ctx) {
+    return ctx->header.n_kv;
+}
+
+int gguf_find_key(struct gguf_context * ctx, const char * key) {
+    // return -1 if key not found
+    int keyfound = -1;
+
+    const int n_kv = gguf_get_n_kv(ctx);
+
+    for (int i = 0; i < n_kv; ++i) {
+        if (strcmp(key, gguf_get_key(ctx, i)) == 0) {
+            keyfound = i;
+            break;
+        }
+    }
+
+    return keyfound;
+}
+
+const char * gguf_get_key(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].key.data;
+}
+
+enum gguf_type gguf_get_kv_type(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].type;
+}
+
+enum gguf_type gguf_get_arr_type(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.arr.type;
+}
+
+const void * gguf_get_arr_data(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.arr.data;
+}
+
+const char * gguf_get_arr_str(struct gguf_context * ctx, int key_id, int i) {
+    struct gguf_kv * kv = &ctx->kv[key_id];
+    struct gguf_str * str = &((struct gguf_str *) kv->value.arr.data)[i];
+    return str->data;
+}
+
+int gguf_get_arr_n(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.arr.n;
+}
+
+uint8_t gguf_get_val_u8(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.uint8;
+}
+
+int8_t gguf_get_val_i8(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.int8;
+}
+
+uint16_t gguf_get_val_u16(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.uint16;
+}
+
+int16_t gguf_get_val_i16(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.int16;
+}
+
+uint32_t gguf_get_val_u32(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.uint32;
+}
+
+int32_t gguf_get_val_i32(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.int32;
+}
+
+float gguf_get_val_f32(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.float32;
+}
+
+bool gguf_get_val_bool(struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.bool_;
+}
+
+const char * gguf_get_val_str (struct gguf_context * ctx, int i) {
+    return ctx->kv[i].value.str.data;
+}
+
+int gguf_get_n_tensors(struct gguf_context * ctx) {
+    return ctx->header.n_tensors;
+}
+
+int gguf_find_tensor(struct gguf_context * ctx, const char * name) {
+    // return -1 if tensor not found
+    int tensorfound = -1;
+
+    const int n_tensors = gguf_get_n_tensors(ctx);
+
+    for (int i = 0; i < n_tensors; ++i) {
+        if (strcmp(name, gguf_get_tensor_name(ctx, i)) == 0) {
+            tensorfound = i;
+            break;
+        }
+    }
+
+    return tensorfound;
+}
+
+size_t gguf_get_tensor_offset(struct gguf_context * ctx, int i) {
+    return ctx->infos[i].offset;
+}
+
+char * gguf_get_tensor_name(struct gguf_context * ctx, int i) {
+    return ctx->infos[i].name.data;
+}
+
+// returns the index
+static int gguf_get_or_add_key(struct gguf_context * ctx, const char * key) {
+    const int idx = gguf_find_key(ctx, key);
+    if (idx >= 0) {
+        return idx;
+    }
+
+    const int n_kv = gguf_get_n_kv(ctx);
+
+    ctx->kv = realloc(ctx->kv, (n_kv + 1) * sizeof(struct gguf_kv));
+    ctx->kv[n_kv].key.n    = strlen(key) + 1;
+    ctx->kv[n_kv].key.data = strdup(key);
+    ctx->header.n_kv++;
+
+    return n_kv;
+}
+
+void gguf_set_val_u8(struct gguf_context * ctx, const char * key, uint8_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_UINT8;
+    ctx->kv[idx].value.uint8 = val;
+}
+
+void gguf_set_val_i8(struct gguf_context * ctx, const char * key, int8_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type       = GGUF_TYPE_INT8;
+    ctx->kv[idx].value.int8 = val;
+}
+
+void gguf_set_val_u16(struct gguf_context * ctx, const char * key, uint16_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type         = GGUF_TYPE_UINT16;
+    ctx->kv[idx].value.uint16 = val;
+}
+
+void gguf_set_val_i16(struct gguf_context * ctx, const char * key, int16_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_INT16;
+    ctx->kv[idx].value.int16 = val;
+}
+
+void gguf_set_val_u32(struct gguf_context * ctx, const char * key, uint32_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type         = GGUF_TYPE_UINT32;
+    ctx->kv[idx].value.uint32 = val;
+}
+
+void gguf_set_val_i32(struct gguf_context * ctx, const char * key, int32_t val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_INT32;
+    ctx->kv[idx].value.int32 = val;
+}
+
+void gguf_set_val_f32(struct gguf_context * ctx, const char * key, float val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type          = GGUF_TYPE_FLOAT32;
+    ctx->kv[idx].value.float32 = val;
+}
+
+void gguf_set_val_bool(struct gguf_context * ctx, const char * key, bool val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type        = GGUF_TYPE_BOOL;
+    ctx->kv[idx].value.bool_ = val;
+}
+
+void gguf_set_val_str(struct gguf_context * ctx, const char * key, const char * val) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_STRING;
+    ctx->kv[idx].value.str.n    = strlen(val) + 1;
+    ctx->kv[idx].value.str.data = strdup(val);
+}
+
+void gguf_set_arr_data(struct gguf_context * ctx, const char * key, enum gguf_type type, const void * data, int n) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_ARRAY;
+    ctx->kv[idx].value.arr.type = type;
+    ctx->kv[idx].value.arr.n    = n;
+    ctx->kv[idx].value.arr.data = malloc(n*GGUF_TYPE_SIZE[type]);
+    memcpy(ctx->kv[idx].value.arr.data, data, n*GGUF_TYPE_SIZE[type]);
+}
+
+void gguf_set_arr_str(struct gguf_context * ctx, const char * key, const char ** data, int n) {
+    const int idx = gguf_get_or_add_key(ctx, key);
+
+    ctx->kv[idx].type           = GGUF_TYPE_ARRAY;
+    ctx->kv[idx].value.arr.type = GGUF_TYPE_STRING;
+    ctx->kv[idx].value.arr.n    = n;
+    ctx->kv[idx].value.arr.data = malloc(n*sizeof(struct gguf_str));
+    for (int i = 0; i < n; i++) {
+        struct gguf_str * str = &((struct gguf_str *)ctx->kv[idx].value.arr.data)[i];
+        str->n    = strlen(data[i]) + 1;
+        str->data = strdup(data[i]);
+    }
+}
+
+// set or add KV pairs from another context
+void gguf_set_kv(struct gguf_context * ctx, struct gguf_context * src) {
+    for (uint32_t i = 0; i < src->header.n_kv; i++) {
+        switch (src->kv[i].type) {
+            case GGUF_TYPE_UINT8:   gguf_set_val_u8  (ctx, src->kv[i].key.data, src->kv[i].value.uint8);    break;
+            case GGUF_TYPE_INT8:    gguf_set_val_i8  (ctx, src->kv[i].key.data, src->kv[i].value.int8);     break;
+            case GGUF_TYPE_UINT16:  gguf_set_val_u16 (ctx, src->kv[i].key.data, src->kv[i].value.uint16);   break;
+            case GGUF_TYPE_INT16:   gguf_set_val_i16 (ctx, src->kv[i].key.data, src->kv[i].value.int16);    break;
+            case GGUF_TYPE_UINT32:  gguf_set_val_u32 (ctx, src->kv[i].key.data, src->kv[i].value.uint32);   break;
+            case GGUF_TYPE_INT32:   gguf_set_val_i32 (ctx, src->kv[i].key.data, src->kv[i].value.int32);    break;
+            case GGUF_TYPE_FLOAT32: gguf_set_val_f32 (ctx, src->kv[i].key.data, src->kv[i].value.float32);  break;
+            case GGUF_TYPE_BOOL:    gguf_set_val_bool(ctx, src->kv[i].key.data, src->kv[i].value.bool_);    break;
+            case GGUF_TYPE_STRING:  gguf_set_val_str (ctx, src->kv[i].key.data, src->kv[i].value.str.data); break;
+            case GGUF_TYPE_ARRAY:
+                {
+                    if (src->kv[i].value.arr.type == GGUF_TYPE_STRING) {
+                        const char ** data = malloc(src->kv[i].value.arr.n*sizeof(char *));
+                        for (uint32_t j = 0; j < src->kv[i].value.arr.n; j++) {
+                            data[j] = ((struct gguf_str *)src->kv[i].value.arr.data)[j].data;
+                        }
+                        gguf_set_arr_str(ctx, src->kv[i].key.data, data, src->kv[i].value.arr.n);
+                        free(data);
+                    } else if (src->kv[i].value.arr.type == GGUF_TYPE_ARRAY) {
+                        GGML_ASSERT(false && "nested arrays not supported");
+                    } else {
+                        gguf_set_arr_data(ctx, src->kv[i].key.data, src->kv[i].value.arr.type, src->kv[i].value.arr.data, src->kv[i].value.arr.n);
+                    }
+                } break;
+            case GGUF_TYPE_COUNT:  GGML_ASSERT(false && "invalid type"); break;
+        }
+    }
+}
+
+void gguf_add_tensor(
+             struct gguf_context * ctx,
+        const struct ggml_tensor * tensor) {
+    const int idx = ctx->header.n_tensors;
+    ctx->infos = realloc(ctx->infos, (idx + 1)*sizeof(struct gguf_tensor_info));
+
+    ctx->infos[idx].name.n    = strlen(tensor->name) + 1;
+    ctx->infos[idx].name.data = strdup(tensor->name);
+
+    for (int i = 0; i < GGML_MAX_DIMS; ++i) {
+        ctx->infos[idx].ne[i] = 1;
+    }
+
+    ctx->infos[idx].n_dims = tensor->n_dims;
+    for (int i = 0; i < tensor->n_dims; i++) {
+        ctx->infos[idx].ne[i] = tensor->ne[i];
+    }
+
+    ctx->infos[idx].type   = tensor->type;
+    ctx->infos[idx].offset = 0;
+    ctx->infos[idx].data   = tensor->data;
+    ctx->infos[idx].size   = ggml_nbytes(tensor);
+
+    if (ctx->header.n_tensors > 0) {
+        ctx->infos[idx].offset = ctx->infos[idx - 1].offset + GGML_PAD(ctx->infos[idx - 1].size, ctx->alignment);
+    }
+
+    ctx->header.n_tensors++;
+}
+
+void gguf_set_tensor_type(struct gguf_context * ctx, const char * name, enum ggml_type type) {
+    const int idx = gguf_find_tensor(ctx, name);
+    if (idx < 0) {
+        GGML_ASSERT(false && "tensor not found");
+    }
+
+    ctx->infos[idx].type = type;
+}
+
+void gguf_set_tensor_data(struct gguf_context * ctx, const char * name, const void * data, size_t size) {
+    const int idx = gguf_find_tensor(ctx, name);
+    if (idx < 0) {
+        GGML_ASSERT(false && "tensor not found");
+    }
+
+    ctx->infos[idx].data = data;
+    ctx->infos[idx].size = size;
+
+    // update offsets
+    for (uint32_t i = idx + 1; i < ctx->header.n_tensors; ++i) {
+        ctx->infos[i].offset = ctx->infos[i - 1].offset + GGML_PAD(ctx->infos[i - 1].size, ctx->alignment);
+    }
+}
+
+//static void gguf_fwrite_str(FILE * file, const struct gguf_str * val) {
+//    fwrite(&val->n,   sizeof(val->n),    1, file);
+//    fwrite(val->data, sizeof(char), val->n, file);
+//}
+//
+//static void gguf_fwrite_el(FILE * file, const void * val, size_t size) {
+//    fwrite(val, sizeof(char), size, file);
+//}
+
+struct gguf_buf {
+    void * data;
+    size_t size;
+    size_t offset;
+};
+
+static struct gguf_buf gguf_buf_init(size_t size) {
+    struct gguf_buf buf = {
+        /*buf.data   =*/ size == 0 ? NULL : malloc(size),
+        /*buf.size   =*/ size,
+        /*buf.offset =*/ 0,
+    };
+
+    return buf;
+}
+
+static void gguf_buf_free(struct gguf_buf buf) {
+    if (buf.data) {
+        free(buf.data);
+    }
+}
+
+static void gguf_buf_grow(struct gguf_buf * buf, size_t size) {
+    if (buf->offset + size > buf->size) {
+        buf->size = 1.5*(buf->offset + size);
+        if (buf->data) {
+            buf->data = realloc(buf->data, buf->size);
+        }
+    }
+}
+
+static void gguf_bwrite_str(struct gguf_buf * buf, const struct gguf_str * val) {
+    gguf_buf_grow(buf, sizeof(val->n) + val->n);
+
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, &val->n, sizeof(val->n));
+    }
+    buf->offset += sizeof(val->n);
+
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, val->data, val->n);
+    }
+    buf->offset += val->n;
+}
+
+static void gguf_bwrite_el(struct gguf_buf * buf, const void * val, size_t el_size) {
+    gguf_buf_grow(buf, el_size);
+
+    if (buf->data) {
+        memcpy((char *) buf->data + buf->offset, val, el_size);
+    }
+    buf->offset += el_size;
+}
+
+static void gguf_write_to_buf(struct gguf_context * ctx, struct gguf_buf * buf, bool only_meta) {
+    // write header
+    gguf_bwrite_el(buf, &ctx->header.magic,     sizeof(ctx->header.magic));
+    gguf_bwrite_el(buf, &ctx->header.version,   sizeof(ctx->header.version));
+    gguf_bwrite_el(buf, &ctx->header.n_tensors, sizeof(ctx->header.n_tensors));
+    gguf_bwrite_el(buf, &ctx->header.n_kv,      sizeof(ctx->header.n_kv));
+
+    // write key-value pairs
+    for (uint32_t i = 0; i < ctx->header.n_kv; ++i) {
+        struct gguf_kv * kv = &ctx->kv[i];
+
+        gguf_bwrite_str(buf, &kv->key);
+        gguf_bwrite_el (buf, &kv->type, sizeof(kv->type));
+
+        switch (kv->type) {
+            case GGUF_TYPE_UINT8:   gguf_bwrite_el( buf, &kv->value.uint8,   sizeof(kv->value.uint8)  ); break;
+            case GGUF_TYPE_INT8:    gguf_bwrite_el (buf, &kv->value.int8,    sizeof(kv->value.int8)   ); break;
+            case GGUF_TYPE_UINT16:  gguf_bwrite_el (buf, &kv->value.uint16,  sizeof(kv->value.uint16) ); break;
+            case GGUF_TYPE_INT16:   gguf_bwrite_el (buf, &kv->value.int16,   sizeof(kv->value.int16)  ); break;
+            case GGUF_TYPE_UINT32:  gguf_bwrite_el (buf, &kv->value.uint32,  sizeof(kv->value.uint32) ); break;
+            case GGUF_TYPE_INT32:   gguf_bwrite_el (buf, &kv->value.int32,   sizeof(kv->value.int32)  ); break;
+            case GGUF_TYPE_FLOAT32: gguf_bwrite_el (buf, &kv->value.float32, sizeof(kv->value.float32)); break;
+            case GGUF_TYPE_BOOL:    gguf_bwrite_el (buf, &kv->value.bool_,   sizeof(kv->value.bool_)  ); break;
+            case GGUF_TYPE_STRING:  gguf_bwrite_str(buf, &kv->value.str                               ); break;
+            case GGUF_TYPE_ARRAY:
+                {
+                    gguf_bwrite_el(buf, &kv->value.arr.type, sizeof(kv->value.arr.type));
+                    gguf_bwrite_el(buf, &kv->value.arr.n,    sizeof(kv->value.arr.n)   );
+
+                    switch (kv->value.arr.type) {
+                        case GGUF_TYPE_UINT8:
+                        case GGUF_TYPE_INT8:
+                        case GGUF_TYPE_UINT16:
+                        case GGUF_TYPE_INT16:
+                        case GGUF_TYPE_UINT32:
+                        case GGUF_TYPE_INT32:
+                        case GGUF_TYPE_FLOAT32:
+                        case GGUF_TYPE_BOOL:
+                            {
+                                gguf_bwrite_el(buf, kv->value.arr.data, kv->value.arr.n * GGUF_TYPE_SIZE[kv->value.arr.type]);
+                            } break;
+                        case GGUF_TYPE_STRING:
+                            {
+                                for (uint32_t j = 0; j < kv->value.arr.n; ++j) {
+                                    gguf_bwrite_str(buf, &((struct gguf_str *) kv->value.arr.data)[j]);
+                                }
+                            } break;
+                        case GGUF_TYPE_ARRAY:
+                        case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type"); break;
+                    };
+                } break;
+            case GGUF_TYPE_COUNT: GGML_ASSERT(false && "invalid type");
+        };
+    }
+
+    // write tensor infos
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        gguf_bwrite_str(buf, &info->name);
+        gguf_bwrite_el (buf, &info->n_dims, sizeof(info->n_dims));
+        for (uint32_t j = 0; j < info->n_dims; ++j) {
+            gguf_bwrite_el(buf, &info->ne[j], sizeof(info->ne[j]));
+        }
+        gguf_bwrite_el(buf, &info->type,   sizeof(info->type));
+        gguf_bwrite_el(buf, &info->offset, sizeof(info->offset));
+    }
+
+    // we require the data section to be aligned, so take into account any padding
+    {
+        const size_t offset     = buf->offset;
+        const size_t offset_pad = GGML_PAD(offset, ctx->alignment);
+
+        if (offset_pad != offset) {
+            uint8_t pad = 0;
+            for (size_t i = 0; i < offset_pad - offset; ++i) {
+                gguf_bwrite_el(buf, &pad, sizeof(pad));
+            }
+        }
+    }
+
+    if (only_meta) {
+        return;
+    }
+
+    size_t offset = 0;
+
+    // write tensor data
+    for (uint32_t i = 0; i < ctx->header.n_tensors; ++i) {
+        struct gguf_tensor_info * info = &ctx->infos[i];
+
+        const size_t size     = info->size;
+        const size_t size_pad = GGML_PAD(size, ctx->alignment);
+
+        gguf_bwrite_el(buf, info->data, size);
+
+        if (size_pad != size) {
+            uint8_t pad = 0;
+            for (size_t j = 0; j < size_pad - size; ++j) {
+                gguf_bwrite_el(buf, &pad, sizeof(pad));
+            }
+        }
+
+        GGML_ASSERT(offset == info->offset);
+
+        offset += size_pad;
+    }
+}
+
+void gguf_write_to_file(struct gguf_context * ctx, const char * fname, bool only_meta) {
+    FILE * file = fopen(fname, "wb");
+    if (!file) {
+        GGML_ASSERT(false && "failed to open file for writing");
+    }
+
+    struct gguf_buf buf = gguf_buf_init(16*1024);
+
+    gguf_write_to_buf(ctx, &buf, only_meta);
+
+    fwrite(buf.data, 1, buf.offset, file);
+
+    gguf_buf_free(buf);
+
+    fclose(file);
+}
+
+size_t gguf_get_meta_size(struct gguf_context * ctx) {
+    // no allocs - only compute size
+    struct gguf_buf buf = gguf_buf_init(0);
+
+    gguf_write_to_buf(ctx, &buf, true);
+
+    return buf.offset;
+}
+
+void gguf_get_meta_data(struct gguf_context * ctx, void * data) {
+    struct gguf_buf buf = gguf_buf_init(16*1024);
+
+    gguf_write_to_buf(ctx, &buf, true);
+
+    memcpy(data, buf.data, buf.offset);
+
+    gguf_buf_free(buf);
+}
+
+////////////////////////////////////////////////////////////////////////////////
+
 int ggml_cpu_has_avx(void) {
 #if defined(__AVX__)
     return 1;