Commit graph

23 commits

Author SHA1 Message Date
comex
723dac55fa
py : new conversion script (#545)
Current status: Working, except for the latest GPTQ-for-LLaMa format
  that includes `g_idx`.  This turns out to require changes to GGML, so
  for now it only works if you use the `--outtype` option to dequantize it
  back to f16 (which is pointless except for debugging).

  I also included some cleanup for the C++ code.

  This script is meant to replace all the existing conversion scripts
  (including the ones that convert from older GGML formats), while also
  adding support for some new formats.  Specifically, I've tested with:

  - [x] `LLaMA` (original)
  - [x] `llama-65b-4bit`
  - [x] `alpaca-native`
  - [x] `alpaca-native-4bit`
  - [x] LLaMA converted to 'transformers' format using
        `convert_llama_weights_to_hf.py`
  - [x] `alpaca-native` quantized with `--true-sequential --act-order
        --groupsize 128` (dequantized only)
  - [x] same as above plus `--save_safetensors`
  - [x] GPT4All
  - [x] stock unversioned ggml
  - [x] ggmh

  There's enough overlap in the logic needed to handle these different
  cases that it seemed best to move to a single script.

  I haven't tried this with Alpaca-LoRA because I don't know where to find
  it.

  Useful features:

  - Uses multiple threads for a speedup in some cases (though the Python
    GIL limits the gain, and sometimes it's disk-bound anyway).

  - Combines split models into a single file (both the intra-tensor split
    of the original and the inter-tensor split of 'transformers' format
    files).  Single files are more convenient to work with and more
    friendly to future changes to use memory mapping on the C++ side.  To
    accomplish this without increasing memory requirements, it has some
    custom loading code which avoids loading whole input files into memory
    at once.

  - Because of the custom loading code, it no longer depends in PyTorch,
    which might make installing dependencies slightly easier or faster...
    although it still depends on NumPy and sentencepiece, so I don't know
    if there's any meaningful difference.  In any case, I also added a
    requirements.txt file to lock the dependency versions in case of any
    future breaking changes.

  - Type annotations checked with mypy.

  - Some attempts to be extra user-friendly:

      - The script tries to be forgiving with arguments, e.g. you can
        specify either the model file itself or the directory containing
        it.

      - The script doesn't depend on config.json / params.json, just in
        case the user downloaded files individually and doesn't have those
        handy.  But you still need tokenizer.model and, for Alpaca,
        added_tokens.json.

      - The script tries to give a helpful error message if
        added_tokens.json is missing.
2023-04-14 10:03:03 +03:00
Pavol Rusnak
cbef542879 py : cleanup the code
- use f-strings where possible
- drop first param of encode/decode functions since "utf-8" is the default
2023-03-31 10:32:01 +02:00
Justine Tunney
ee0c40dd6d Introduce GGML migration tool for new file format
If you deleted your old Meta LLaMA .pth files, then the
migrate-ggml-2023-03-30-pr613.py script will allow you to convert your
old ggml files into the new mmap()'able format.

See #613
2023-03-30 12:28:25 -07:00
Justine Tunney
78ca9838ee Make loading weights 10-100x faster
This is a breaking change that's going to give you three benefits:

1. Your inference commands should load 100x faster
2. You may be able to safely load models 2x larger
3. You can run many concurrent inference processes

This was accomplished by changing the file format so we can mmap()
weights directly into memory without having to read() or copy them
thereby ensuring the kernel can make its file cache pages directly
accessible to our inference processes; and secondly, that the file
cache pages are much less likely to get evicted (which would force
loads to hit disk) because they're no longer competing with memory
pages that were needlessly created by gigabytes of standard i/o.

The new file format supports single-file models like LLaMA 7b, and
it also supports multi-file models like LLaMA 13B. Our Python tool
now merges the foo.1, foo.2, etc. files back into a single file so
that the C++ code which maps it doesn't need to reshape data every
time. That's made llama.cpp so much simpler. Much of its load code
has now been deleted.

Furthermore, this change ensures that tensors are aligned properly
on a 32-byte boundary. That opens the door to seeing if we can get
additional performance gains on some microprocessors, by using ops
that require memory alignment.

Lastly note that both POSIX and the Windows platform are supported

Fixes #91
2023-03-30 12:28:25 -07:00
DooWoong Lee (David)
692ce3164e
py : removed unused model variable and verified that the code functions correctly with vocab_only setting. Also confirmed that the code works as expected after running with reduced memory usage due to deletion of no-longer-needed variable. (#547) 2023-03-28 20:02:34 +03:00
jp-x-g
f732695cd5
Clarify console output in convert-pth-to-ggml.py (#512)
"Processing part 1 of 3" instead of "Processing part 0"
2023-03-25 23:53:55 +02:00
Georgi Gerganov
f5a77a629b
Introduce C-style API (#370)
* Major refactoring - introduce C-style API

* Clean up

* Add <cassert>

* Add <iterator>

* Add <algorithm> ....

* Fix timing reporting and accumulation

* Measure eval time only for single-token calls

* Change llama_tokenize return meaning
2023-03-22 07:32:36 +02:00
Georgi Gerganov
3bfa3b43b7
Fix convert script, warnings alpaca instructions, default params 2023-03-21 17:59:16 +02:00
Mack Straight
c98ae02668
fix typo in comment (#318) 2023-03-21 17:49:43 +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
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
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
Georgi Gerganov
c1c7026b47
Fix python stuff (#109) 2023-03-19 19:33:18 +02:00
qunash
467b149761
Refactoring convert-pth-to-ggml.py: more concise and readable (#109)
* Refactor get_n_parts function to simplify code and improve readability

* Use f-strings instead of concatenation

* Refactoring: more concise and readable

* modularize

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-03-19 19:17:39 +02:00
Bernat Vadell
2af23d3043
🚀 Dockerize llamacpp (#132)
* feat: dockerize llamacpp

* feat: split build & runtime stages

* split dockerfile into main & tools

* add quantize into tool docker image

* Update .devops/tools.sh

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

* add docker action pipeline

* change CI to publish at github docker registry

* fix name runs-on macOS-latest is macos-latest (lowercase)

* include docker versioned images

* fix github action docker

* fix docker.yml

* feat: include all-in-one command tool & update readme.md

---------

Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
2023-03-17 10:47:06 +01:00
Ronsor
956dfda8ad
Use tokenizer.vocab_size() instead of hardcoding 32000 in convert-pth-to-ggml.py (#142)
There are ways that special tokens or other new tokens could be added to the tokenizer; therefore it's probably best not to assume the vocabulary is only 32000 tokens.
2023-03-15 21:37:50 +02:00
Val Kharitonov
2a20f48efa
Fix UTF-8 handling (including colors) (#79) 2023-03-13 18:24:18 +02:00
Georgi Gerganov
7c9e54e55e
Revert "weights_only" arg - this causing more trouble than help 2023-03-12 20:59:01 +02:00
Oleksandr Nikitin
b9bd1d0141
python/pytorch compat notes (#44) 2023-03-12 14:16:33 +02:00
deepdiffuser
a93120236f
use weights_only in conversion script (#32)
this restricts malicious weights from executing arbitrary code by restricting the unpickler to only loading tensors, primitive types, and dictionaries
2023-03-12 08:36:35 +02:00
Georgi Gerganov
007a8f6f45
Support all LLaMA models + change Q4_0 quantization storage 2023-03-11 11:28:30 +02:00
Georgi Gerganov
70bc0b8b15
Fix a bug in the rope calculation 2023-03-10 23:46:57 +02:00
Georgi Gerganov
26c0846629
Initial release 2023-03-10 20:56:40 +02:00