CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-11-02 09:12:03 +00:00
Code Issues Packages Projects Releases Wiki Activity

convert-*.py: GGUF Naming Convention Refactor and Metadata Override Refactor (#7499)

Browse Source 
Main thing is that the default output filename will take this form

{name}{parameters}{finetune}{version}{encoding}{kind}

In addition this add and remove some entries in the KV store and adds a metadata class with automatic heuristics capability to derive some values based on model card content

* No Change:
  - Internal GGUF Spec
    - `general.architecture`
    - `general.quantization_version`
    - `general.alignment`
    - `general.file_type`
  - General Model Details
    - `general.name`
    - `general.author`
    - `general.version`
    - `general.description`
  - Licensing details
    - `general.license`
  - Typically represents the converted GGUF repo (Unless made from scratch)
    - `general.url`
  - Model Source during conversion
    - `general.source.url`

* Removed:
  - Model Source during conversion
    - `general.source.huggingface.repository`

* Added:
  - General Model Details
    - `general.organization`
    - `general.finetune`
    - `general.basename`
    - `general.quantized_by`
    - `general.size_label`
  - Licensing details
    - `general.license.name`
    - `general.license.link`
  - Typically represents the converted GGUF repo (Unless made from scratch)
    - `general.doi`
    - `general.uuid`
    - `general.repo_url`
  - Model Source during conversion
    - `general.source.doi`
    - `general.source.uuid`
    - `general.source.repo_url`
  - Base Model Source
    - `general.base_model.count`
    - `general.base_model.{id}.name`
    - `general.base_model.{id}.author`
    - `general.base_model.{id}.version`
    - `general.base_model.{id}.organization`
    - `general.base_model.{id}.url` (Model Website/Paper)
    - `general.base_model.{id}.doi`
    - `general.base_model.{id}.uuid`
    - `general.base_model.{id}.repo_url` (Model Source Repository (git/svn/etc...))
  - Array based KV stores
    - `general.tags`
    - `general.languages`
    - `general.datasets`

---------

Co-authored-by: compilade <git@compilade.net>
Co-authored-by: Xuan Son Nguyen <thichthat@gmail.com>
This commit is contained in:
Brian
2024-07-18 20:40:15 +10:00
committed by GitHub
parent 3807c3de04
commit 672a6f1018
13 changed files with 1185 additions and 239 deletions
Download Patch File Download Diff File Expand all files Collapse all files

29
convert_lora_to_gguf.py
View File

@@ -251,6 +251,10 @@ def parse_args() -> argparse.Namespace:
"--verbose", action="store_true",
help="increase output verbosity",
)
parser.add_argument(
"--dry-run", action="store_true",
help="only print out what will be done, without writing any new files",
)
parser.add_argument(
"--base", type=Path, required=True,
help="directory containing base model file",
@@ -300,6 +304,12 @@ if __name__ == '__main__':
# load base model
logger.info(f"Loading base model: {dir_base_model.name}")
hparams = Model.load_hparams(dir_base_model)
with open(lora_config, "r") as f:
lparams: dict[str, Any] = json.load(f)
alpha: float = lparams["lora_alpha"]
with torch.inference_mode():
try:
model_class = Model.from_model_architecture(hparams["architectures"][0])
@@ -310,6 +320,14 @@ if __name__ == '__main__':
class LoraModel(model_class):
model_arch = model_class.model_arch
def set_type(self):
self.gguf_writer.add_type(gguf.GGUFType.ADAPTER)
self.gguf_writer.add_string(gguf.Keys.Adapter.TYPE, "lora")
def set_gguf_parameters(self):
self.gguf_writer.add_float32(gguf.Keys.Adapter.LORA_ALPHA, float(alpha))
super().set_gguf_parameters()
def get_tensors(self) -> Iterator[tuple[str, Tensor]]:
tensor_map: dict[str, PartialLoraTensor] = {}
@@ -357,18 +375,9 @@ if __name__ == '__main__':
is_big_endian=args.bigendian,
use_temp_file=False,
eager=args.no_lazy,
model_name=None,
dry_run=args.dry_run,
)
with open(lora_config, "r") as f:
lparams: dict[str, Any] = json.load(f)
alpha = lparams["lora_alpha"]
model_instance.gguf_writer.add_string(gguf.Keys.General.TYPE, gguf.GGUFType.ADAPTER)
model_instance.gguf_writer.add_string(gguf.Keys.Adapter.TYPE, "lora")
model_instance.gguf_writer.add_float32(gguf.Keys.Adapter.LORA_ALPHA, float(alpha))
model_instance.gguf_writer.add_quantization_version(gguf.GGML_QUANT_VERSION)
logger.info("Exporting model...")
model_instance.write()
logger.info(f"Model successfully exported to {model_instance.fname_out}")