Skip to content

convert-hf : set the model name based on cli arg, if present #7693

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jun 9, 2024
Merged

convert-hf : set the model name based on cli arg, if present #7693

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
ed6f58c
convert-hf : set the model name based on cli arg, if present
sasha0552 Jun 2, 2024
83e1c15
Merge branch 'master' into convert-model-name
mofosyne Jun 9, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
26 changes: 14 additions & 12 deletions convert-hf-to-gguf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,6 +52,7 @@ class Model:
endianess: gguf.GGUFEndian
use_temp_file: bool
lazy: bool
model_name: str | None
part_names: list[str]
is_safetensors: bool
hparams: dict[str, Any]
Expand All @@ -64,7 +65,7 @@ class Model:
# subclasses should define this!
model_arch: gguf.MODEL_ARCH

def __init__(self, dir_model: Path, ftype: gguf.LlamaFileType, fname_out: Path, is_big_endian: bool, use_temp_file: bool, eager: bool):
def __init__(self, dir_model: Path, ftype: gguf.LlamaFileType, fname_out: Path, is_big_endian: bool, use_temp_file: bool, eager: bool, model_name: str | None):
if type(self) is Model:
raise TypeError(f"{type(self).__name__!r} should not be directly instantiated")
self.dir_model = dir_model
Expand All @@ -73,6 +74,7 @@ def __init__(self, dir_model: Path, ftype: gguf.LlamaFileType, fname_out: Path,
self.endianess = gguf.GGUFEndian.BIG if is_big_endian else gguf.GGUFEndian.LITTLE
self.use_temp_file = use_temp_file
self.lazy = not eager
self.model_name = model_name
self.part_names = Model.get_model_part_names(self.dir_model, "model", ".safetensors")
self.is_safetensors = len(self.part_names) > 0
if not self.is_safetensors:
Expand Down Expand Up @@ -182,7 +184,7 @@ def map_tensor_name(self, name: str, try_suffixes: Sequence[str] = (".weight", "
return new_name

def set_gguf_parameters(self):
self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_block_count(self.block_count)

if (n_ctx := self.find_hparam(["max_position_embeddings", "n_ctx"], optional=True)) is not None:
Expand Down Expand Up @@ -665,7 +667,7 @@ class GPTNeoXModel(Model):
def set_gguf_parameters(self):
block_count = self.hparams["num_hidden_layers"]

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_context_length(self.hparams["max_position_embeddings"])
self.gguf_writer.add_embedding_length(self.hparams["hidden_size"])
self.gguf_writer.add_block_count(block_count)
Expand Down Expand Up @@ -798,7 +800,7 @@ def set_vocab(self):

def set_gguf_parameters(self):
block_count = self.hparams["n_layers"]
self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_context_length(self.hparams["max_seq_len"])
self.gguf_writer.add_embedding_length(self.hparams["d_model"])
self.gguf_writer.add_block_count(block_count)
Expand Down Expand Up @@ -850,7 +852,7 @@ def set_gguf_parameters(self):
raise ValueError("gguf: can not find ctx length parameter.")

self.gguf_writer.add_file_type(self.ftype)
self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_source_hf_repo(hf_repo)
self.gguf_writer.add_tensor_data_layout("Meta AI original pth")
self.gguf_writer.add_context_length(ctx_length)
Expand Down Expand Up @@ -887,7 +889,7 @@ def set_gguf_parameters(self):
else:
raise ValueError("gguf: can not find ctx length parameter.")

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_source_hf_repo(hf_repo)
self.gguf_writer.add_tensor_data_layout("Meta AI original pth")
self.gguf_writer.add_context_length(ctx_length)
Expand Down Expand Up @@ -1010,7 +1012,7 @@ def set_gguf_parameters(self):
else:
raise ValueError("gguf: can not find ctx length parameter.")

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_source_hf_repo(hf_repo)
self.gguf_writer.add_tensor_data_layout("Meta AI original pth")
self.gguf_writer.add_context_length(ctx_length)
Expand Down Expand Up @@ -1206,7 +1208,7 @@ def set_gguf_parameters(self):
hparams = self.hparams
block_count = hparams["num_hidden_layers"]

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_context_length(hparams["max_position_embeddings"])
self.gguf_writer.add_embedding_length(hparams["hidden_size"])
self.gguf_writer.add_block_count(block_count)
Expand Down Expand Up @@ -1681,7 +1683,7 @@ class GPT2Model(Model):
model_arch = gguf.MODEL_ARCH.GPT2

def set_gguf_parameters(self):
self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_block_count(self.hparams["n_layer"])
self.gguf_writer.add_context_length(self.hparams["n_ctx"])
self.gguf_writer.add_embedding_length(self.hparams["n_embd"])
Expand Down Expand Up @@ -2248,7 +2250,7 @@ def set_gguf_parameters(self):
hparams = self.hparams
block_count = hparams["num_hidden_layers"]

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_context_length(hparams["max_position_embeddings"])
self.gguf_writer.add_embedding_length(hparams["hidden_size"])
self.gguf_writer.add_block_count(block_count)
Expand Down Expand Up @@ -2348,7 +2350,7 @@ def set_gguf_parameters(self):
# Fail early for models which don't have a block expansion factor of 2
assert d_inner == 2 * d_model

self.gguf_writer.add_name(self.dir_model.name)
self.gguf_writer.add_name(self.dir_model.name if self.model_name is None else self.model_name)
self.gguf_writer.add_context_length(2**20) # arbitrary value; for those who use the default
self.gguf_writer.add_embedding_length(d_model)
self.gguf_writer.add_feed_forward_length(0) # unused, but seemingly required when loading
Expand Down Expand Up @@ -2852,7 +2854,7 @@ def main() -> None:
logger.error(f"Model {hparams['architectures'][0]} is not supported")
sys.exit(1)

model_instance = model_class(dir_model, ftype_map[args.outtype], fname_out, args.bigendian, args.use_temp_file, args.no_lazy)
model_instance = model_class(dir_model, ftype_map[args.outtype], fname_out, args.bigendian, args.use_temp_file, args.no_lazy, args.model_name)

logger.info("Set model parameters")
model_instance.set_gguf_parameters()
Expand Down
Loading