From 6381d4e110bd0ec02843a60bbeb8b6fc37a9ace9 Mon Sep 17 00:00:00 2001 From: Georgi Gerganov Date: Mon, 21 Aug 2023 23:07:43 +0300 Subject: gguf : new file format with flexible meta data (beta) (#2398) MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit * 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 * 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 * 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 * 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 Co-authored-by: klosax <131523366+klosax@users.noreply.github.com> Co-authored-by: goerch Co-authored-by: slaren Co-authored-by: Kerfuffle <44031344+KerfuffleV2@users.noreply.github.com> --- gguf.py | 718 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 718 insertions(+) create mode 100644 gguf.py (limited to 'gguf.py') diff --git a/gguf.py b/gguf.py new file mode 100644 index 00000000..9776649c --- /dev/null +++ b/gguf.py @@ -0,0 +1,718 @@ +import shutil +import sys +import struct +import tempfile +import numpy as np + +from enum import IntEnum, auto +from typing import Any, IO, List, Optional + +# +# constants +# + +GGUF_MAGIC = 0x46554747 +GGUF_VERSION = 1 +GGUF_DEFAULT_ALIGNMENT = 32 + +# general +KEY_GENERAL_ARCHITECTURE = "general.architecture" +KEY_GENERAL_QUANTIZATION_VERSION = "general.quantization_version" +KEY_GENERAL_ALIGNMENT = "general.alignment" +KEY_GENERAL_NAME = "general.name" +KEY_GENERAL_AUTHOR = "general.author" +KEY_GENERAL_URL = "general.url" +KEY_GENERAL_DESCRIPTION = "general.description" +KEY_GENERAL_LICENSE = "general.license" +KEY_GENERAL_SOURCE_URL = "general.source.url" +KEY_GENERAL_SOURCE_HF_REPO = "general.source.hugginface.repository" + +# LLM +KEY_LLM_CONTEXT_LENGTH = "{arch}.context_length" +KEY_LLM_EMBEDDING_LENGTH = "{arch}.embedding_length" +KEY_LLM_BLOCK_COUNT = "{arch}.block_count" +KEY_LLM_FEED_FORWARD_LENGTH = "{arch}.feed_forward_length" +KEY_LLM_USE_PARALLEL_RESIDUAL = "{arch}.use_parallel_residual" +KEY_LLM_TENSOR_DATA_LAYOUT = "{arch}.tensor_data_layout" + +# attention +KEY_ATTENTION_HEAD_COUNT = "{arch}.attention.head_count" +KEY_ATTENTION_HEAD_COUNT_KV = "{arch}.attention.head_count_kv" +KEY_ATTENTION_MAX_ALIBI_BIAS = "{arch}.attention.max_alibi_bias" +KEY_ATTENTION_CLAMP_KQV = "{arch}.attention.clamp_kqv" +KEY_ATTENTION_LAYERNORM_EPS = "{arch}.attention.layer_norm_epsilon" +KEY_ATTENTION_LAYERNORM_RMS_EPS = "{arch}.attention.layer_norm_rms_epsilon" + +# RoPE +KEY_ROPE_DIMENSION_COUNT = "{arch}.rope.dimension_count" +KEY_ROPE_SCALE_LINEAR = "{arch}.rope.scale_linear" + +# tokenization +KEY_TOKENIZER_MODEL = "tokenizer.ggml.model" +KEY_TOKENIZER_LIST = "tokenizer.ggml.tokens" +KEY_TOKENIZER_TOKEN_TYPE = "tokenizer.ggml.token_type" +KEY_TOKENIZER_SCORES = "tokenizer.ggml.scores" +KEY_TOKENIZER_MERGES = "tokenizer.ggml.merges" +KEY_TOKENIZER_BOS_ID = "tokenizer.ggml.bos_token_id" +KEY_TOKENIZER_EOS_ID = "tokenizer.ggml.eos_token_id" +KEY_TOKENIZER_UNK_ID = "tokenizer.ggml.unknown_token_id" +KEY_TOKENIZER_SEP_ID = "tokenizer.ggml.seperator_token_id" +KEY_TOKENIZER_PAD_ID = "tokenizer.ggml.padding_token_id" +KEY_TOKENIZER_HF_JSON = "tokenizer.huggingface.json" +KEY_TOKENIZER_RWKV = "tokenizer.rwkv.world" + + +# +# recommended mapping of model tensor names for storage in gguf +# + + +class MODEL_ARCH(IntEnum): + LLAMA = auto() + FALCON = auto() + GPT2 = auto() + GPTJ = auto() + GPTNEOX = auto() + MPT = auto() + + +class MODEL_TENSOR(IntEnum): + TOKEN_EMBD = auto() + POS_EMBD = auto() + OUTPUT = auto() + OUTPUT_NORM = auto() + ROPE_FREQS = auto() + ATTN_Q = auto() + ATTN_K = auto() + ATTN_V = auto() + ATTN_QKV = auto() + ATTN_OUT = auto() + ATTN_NORM = auto() + ATTN_NORM_2 = auto() + ATTN_ROT_EMBD = auto() + FFN_GATE = auto() + FFN_DOWN = auto() + FFN_UP = auto() + FFN_NORM = auto() + + +MODEL_ARCH_NAMES = { + MODEL_ARCH.LLAMA: "llama", + MODEL_ARCH.FALCON: "falcon", + MODEL_ARCH.GPT2: "gpt2", + MODEL_ARCH.GPTJ: "gptj", + MODEL_ARCH.GPTNEOX: "gptneox", + MODEL_ARCH.MPT: "mpt", +} + +MODEL_TENSOR_NAMES = { + MODEL_ARCH.LLAMA: { + MODEL_TENSOR.TOKEN_EMBD: "token_embd", + MODEL_TENSOR.OUTPUT_NORM: "output_norm", + MODEL_TENSOR.OUTPUT: "output", + MODEL_TENSOR.ROPE_FREQS: "rope_freqs", + MODEL_TENSOR.ATTN_NORM: "blk.{bid}.attn_norm", + MODEL_TENSOR.ATTN_Q: "blk.{bid}.attn_q", + MODEL_TENSOR.ATTN_K: "blk.{bid}.attn_k", + MODEL_TENSOR.ATTN_V: "blk.{bid}.attn_v", + MODEL_TENSOR.ATTN_OUT: "blk.{bid}.attn_output", + MODEL_TENSOR.ATTN_ROT_EMBD: "blk.{bid}.attn_rot_embd", + MODEL_TENSOR.FFN_NORM: "blk.{bid}.ffn_norm", + MODEL_TENSOR.FFN_GATE: "blk.{bid}.ffn_gate", + MODEL_TENSOR.FFN_DOWN: "blk.{bid}.ffn_down", + MODEL_TENSOR.FFN_UP: "blk.{bid}.ffn_up", + }, + MODEL_ARCH.GPTNEOX: { + MODEL_TENSOR.TOKEN_EMBD: "token_embd", + MODEL_TENSOR.OUTPUT_NORM: "output_norm", + MODEL_TENSOR.OUTPUT: "output", + MODEL_TENSOR.ATTN_NORM: "blk.{bid}.attn_norm", + MODEL_TENSOR.ATTN_QKV: "blk.{bid}.attn_qkv", + MODEL_TENSOR.ATTN_OUT: "blk.{bid}.attn_output", + MODEL_TENSOR.FFN_NORM: "blk.{bid}.ffn_norm", + MODEL_TENSOR.FFN_DOWN: "blk.{bid}.ffn_down", + MODEL_TENSOR.FFN_UP: "blk.{bid}.ffn_up", + }, + MODEL_ARCH.FALCON: { + MODEL_TENSOR.TOKEN_EMBD: "token_embd", + MODEL_TENSOR.OUTPUT_NORM: "output_norm", + MODEL_TENSOR.OUTPUT: "output", + MODEL_TENSOR.ATTN_NORM: "blk.{bid}.attn_norm", + MODEL_TENSOR.ATTN_NORM_2: "blk.{bid}.attn_norm_2", + MODEL_TENSOR.ATTN_QKV: "blk.{bid}.attn_qkv", + MODEL_TENSOR.ATTN_OUT: "blk.{bid}.attn_output", + MODEL_TENSOR.FFN_DOWN: "blk.{bid}.ffn_down", + MODEL_TENSOR.FFN_UP: "blk.{bid}.ffn_up", + }, + MODEL_ARCH.GPT2: { + # TODO + }, + # TODO +} + +# tensors that will not be serialized +MODEL_TENSOR_SKIP = { + MODEL_ARCH.LLAMA: [ + MODEL_TENSOR.ROPE_FREQS, + MODEL_TENSOR.ATTN_ROT_EMBD, + ], +} + + +# TODO: the following helper functions should be removed +# instead, get_tensor_name_map should return tuples of (name, MODEL_TENSOR) +# however, my Python is very bad, and I couldn't figure out how to do this, hence these functions +# REMOVE +def should_skip_tensor_TMP(arch: MODEL_ARCH, n_blocks: int, name: str) -> bool: + for skip in MODEL_TENSOR_SKIP.get(arch, []): + for i in range(n_blocks): + if name == MODEL_TENSOR_NAMES[arch][skip].format(bid=i): + return True + + return False + + +def get_tensor_name_map(arch: MODEL_ARCH, n_blocks: int) -> dict: + tensor_map = {} + + # Token embeddings + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.TOKEN_EMBD, None) + + tensor_map["gpt_neox.embed_in"] = mapped_to # gptneox + tensor_map["transformer.wte"] = mapped_to # gpt2 mpt + tensor_map["transformer.word_embeddings"] = mapped_to # falcon + tensor_map["model.embed_tokens"] = mapped_to # llama-hf + tensor_map["tok_embeddings"] = mapped_to # llama-pth + + # Position embeddings + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.POS_EMBD, None) + + tensor_map["transformer.wpe"] = mapped_to # gpt2 + + # Output + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT, None) + + tensor_map["embed_out"] = mapped_to # gptneox + tensor_map["lm_head"] = mapped_to # gpt2 mpt falcon llama-hf + tensor_map["output"] = mapped_to # llama-pth + + # Output norm + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.OUTPUT_NORM, None) + + tensor_map["gpt_neox.final_layer_norm"] = mapped_to # gptneox + tensor_map["transformer.ln_f"] = mapped_to # gpt2 falcon + tensor_map["transformer.norm_f"] = mapped_to # mpt + tensor_map["model.norm"] = mapped_to # llama-hf + tensor_map["norm"] = mapped_to # llama-pth + + # Rope frequencies + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ROPE_FREQS, None) + + tensor_map["rope.freqs"] = mapped_to # llama-pth + + # Attention and feed-forward blocks + for i in range(0, n_blocks): + # Attention norm + # TODO: is there are simpler way to write these 2 lines in Python? + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM, None) + mapped_to = mapped_to.format(bid=i) if mapped_to else None + + tensor_map["gpt_neox.layers."+str(i)+".input_layernorm"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".ln_1"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".norm_1"] = mapped_to # mpt + tensor_map["transformer.h."+str(i)+".input_layernorm"] = mapped_to # falcon7b + tensor_map["transformer.h."+str(i)+".ln_mlp"] = mapped_to # falcon40b + tensor_map["model.layers."+str(i)+".input_layernorm"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention_norm"] = mapped_to # llama-pth + + # Attention norm 2 + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_NORM_2, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["transformer.h."+str(i)+".ln_attn"] = mapped_to # falcon40b + + # Attention query-key-value + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_QKV, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["gpt_neox.layers."+str(i)+".attention.query_key_value"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".attn.c_attn"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".attn.Wqkv"] = mapped_to # mpt + tensor_map["transformer.h."+str(i)+".self_attention.query_key_value"] = mapped_to # falcon + + # Attention query + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_Q, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["model.layers."+str(i)+".self_attn.q_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention.wq"] = mapped_to # llama-pth + + # Attention key + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_K, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["model.layers."+str(i)+".self_attn.k_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention.wk"] = mapped_to # llama-pth + + # Attention value + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_V, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["model.layers."+str(i)+".self_attn.v_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention.wv"] = mapped_to # llama-pth + + # Attention output + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_OUT, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["gpt_neox.layers."+str(i)+".attention.dense"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".attn.c_proj"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".attn.out_proj"] = mapped_to # mpt + tensor_map["transformer.h."+str(i)+".self_attention.dense"] = mapped_to # falcon + tensor_map["model.layers."+str(i)+".self_attn.o_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention.wo"] = mapped_to # llama-pth + + # Rotary embeddings + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.ATTN_ROT_EMBD, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["model.layers."+str(i)+".self_attn.rotary_emb.inv_freq"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".attention.inner_attention.rope.freqs"] = mapped_to # llama-pth + + # Feed-forward norm + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_NORM, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["gpt_neox.layers."+str(i)+".post_attention_layernorm"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".ln_2"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".norm_2"] = mapped_to # mpt + tensor_map["model.layers."+str(i)+".post_attention_layernorm"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".ffn_norm"] = mapped_to # llama-pth + + # Feed-forward up + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_UP, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".mlp.c_fc"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".ffn.up_proj"] = mapped_to # mpt + tensor_map["transformer.h."+str(i)+".mlp.dense_h_to_4h"] = mapped_to # falcon + tensor_map["model.layers."+str(i)+".mlp.up_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".feed_forward.w3"] = mapped_to # llama-pth + + # Feed-forward gate + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_GATE, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["model.layers."+str(i)+".mlp.gate_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".feed_forward.w1"] = mapped_to # llama-pth + + # Feed-forward down + mapped_to = MODEL_TENSOR_NAMES[arch].get(MODEL_TENSOR.FFN_DOWN, None) + mapped_to = mapped_to.format(bid=i) if mapped_to is not None else None + + tensor_map["gpt_neox.layers."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # gptneox + tensor_map["transformer.h."+str(i)+".mlp.c_proj"] = mapped_to # gpt2 + tensor_map["transformer.blocks."+str(i)+".ffn.down_proj"] = mapped_to # mpt + tensor_map["transformer.h."+str(i)+".mlp.dense_4h_to_h"] = mapped_to # falcon + tensor_map["model.layers."+str(i)+".mlp.down_proj"] = mapped_to # llama-hf + tensor_map["layers."+str(i)+".feed_forward.w2"] = mapped_to # llama-pth + + return tensor_map + + +class TokenType(IntEnum): + NORMAL = 1 + UNKNOWN = 2 + CONTROL = 3 + USER_DEFINED = 4 + UNUSED = 5 + BYTE = 6 + +# +# implementation +# + + +class GGMLQuantizationType(IntEnum): + F32 = 0 + F16 = 1 + Q4_0 = 2 + Q4_1 = 3 + Q5_0 = 6 + Q5_1 = 7 + Q8_0 = 8 + Q8_1 = 9 + Q2_K = 10 + Q3_K = 11 + Q4_K = 12 + Q5_K = 13 + Q6_K = 14 + Q8_K = 15 + + +class GGUFValueType(IntEnum): + UINT8 = 0 + INT8 = 1 + UINT16 = 2 + INT16 = 3 + UINT32 = 4 + INT32 = 5 + FLOAT32 = 6 + BOOL = 7 + STRING = 8 + ARRAY = 9 + + @staticmethod + def get_type(val): + if isinstance(val, str) or isinstance(val, bytes) or isinstance(val, bytearray): + return GGUFValueType.STRING + elif isinstance(val, list): + return GGUFValueType.ARRAY + elif isinstance(val, float): + return GGUFValueType.FLOAT32 + elif isinstance(val, bool): + return GGUFValueType.BOOL + elif isinstance(val, int): + return GGUFValueType.INT32 + else: + print("Unknown type: "+str(type(val))) + sys.exit() + + +class GGUFWriter: + def __init__(self, path: str, arch: str, use_temp_file = True): + self.fout = open(path, "wb") + self.arch = arch + self.offset_tensor = 0 + self.data_alignment = GGUF_DEFAULT_ALIGNMENT + self.kv_data = b"" + self.kv_data_count = 0 + self.ti_data = b"" + self.ti_data_count = 0 + self.add_architecture() + self.use_temp_file = use_temp_file + self.tensors = [] + + def write_header_to_file(self): + self.fout.write(struct.pack(" int: + return ((x + n - 1) // n) * n + + def add_tensor_info(self, name: str, tensor_shape: np.ndarray, tensor_dtype: np.dtype, tensor_nbytes: int, raw_dtype: Optional[GGMLQuantizationType] = None): + assert raw_dtype is not None or tensor_dtype in (np.float32, np.float16), "Only F32 and F16 tensors are supported for now" + + encoded_name = name.encode("utf8") + self.ti_data += struct.pack("