Returning the PR as per old code (still passing `api` instead of `token` directly
Browse files- convert.py +14 -7
convert.py
CHANGED
|
@@ -7,9 +7,9 @@ import torch
|
|
| 7 |
|
| 8 |
from huggingface_hub import CommitOperationAdd, HfApi, hf_hub_download
|
| 9 |
from huggingface_hub.file_download import repo_folder_name
|
|
|
|
| 10 |
from transformers import AutoConfig
|
| 11 |
from transformers.pipelines.base import infer_framework_load_model
|
| 12 |
-
from safetensors.torch import save_file
|
| 13 |
|
| 14 |
|
| 15 |
def check_file_size(sf_filename, pt_filename):
|
|
@@ -17,10 +17,12 @@ def check_file_size(sf_filename, pt_filename):
|
|
| 17 |
pt_size = os.stat(pt_filename).st_size
|
| 18 |
|
| 19 |
if (sf_size - pt_size) / pt_size > 0.01:
|
| 20 |
-
raise RuntimeError(
|
|
|
|
| 21 |
- {sf_filename}: {sf_size}
|
| 22 |
- {pt_filename}: {pt_size}
|
| 23 |
-
"""
|
|
|
|
| 24 |
|
| 25 |
|
| 26 |
def rename(pt_filename) -> str:
|
|
@@ -29,12 +31,13 @@ def rename(pt_filename) -> str:
|
|
| 29 |
return local
|
| 30 |
|
| 31 |
|
| 32 |
-
def convert_multi(model_id):
|
| 33 |
filename = hf_hub_download(repo_id=model_id, filename="pytorch_model.bin.index.json")
|
| 34 |
with open(filename, "r") as f:
|
| 35 |
data = json.load(f)
|
| 36 |
|
| 37 |
filenames = set(data["weight_map"].values())
|
|
|
|
| 38 |
for filename in filenames:
|
| 39 |
cached_filename = hf_hub_download(repo_id=model_id, filename=filename)
|
| 40 |
loaded = torch.load(cached_filename)
|
|
@@ -53,7 +56,9 @@ def convert_multi(model_id):
|
|
| 53 |
json.dump(newdata, f)
|
| 54 |
local_filenames.append(index)
|
| 55 |
|
| 56 |
-
operations = [
|
|
|
|
|
|
|
| 57 |
|
| 58 |
return operations
|
| 59 |
|
|
@@ -71,6 +76,7 @@ def convert_single(model_id, folder):
|
|
| 71 |
operations = [CommitOperationAdd(path_in_repo=sf_filename, path_or_fileobj=local)]
|
| 72 |
return operations
|
| 73 |
|
|
|
|
| 74 |
def check_final_model(model_id, folder):
|
| 75 |
config = hf_hub_download(repo_id=model_id, filename="config.json")
|
| 76 |
shutil.copy(config, os.path.join(folder, "config.json"))
|
|
@@ -91,6 +97,7 @@ def convert(api, model_id):
|
|
| 91 |
|
| 92 |
folder = repo_folder_name(repo_id=model_id, repo_type="models")
|
| 93 |
os.makedirs(folder)
|
|
|
|
| 94 |
try:
|
| 95 |
operations = None
|
| 96 |
if "model.safetensors" in filenames or "model_index.safetensors.index.json" in filenames:
|
|
@@ -104,7 +111,7 @@ def convert(api, model_id):
|
|
| 104 |
|
| 105 |
if operations:
|
| 106 |
check_final_model(model_id, folder)
|
| 107 |
-
api.create_commit(
|
| 108 |
repo_id=model_id,
|
| 109 |
operations=operations,
|
| 110 |
commit_message="Adding `safetensors` variant of this model",
|
|
@@ -112,7 +119,7 @@ def convert(api, model_id):
|
|
| 112 |
)
|
| 113 |
finally:
|
| 114 |
shutil.rmtree(folder)
|
| 115 |
-
return
|
| 116 |
|
| 117 |
|
| 118 |
if __name__ == "__main__":
|
|
|
|
| 7 |
|
| 8 |
from huggingface_hub import CommitOperationAdd, HfApi, hf_hub_download
|
| 9 |
from huggingface_hub.file_download import repo_folder_name
|
| 10 |
+
from safetensors.torch import save_file
|
| 11 |
from transformers import AutoConfig
|
| 12 |
from transformers.pipelines.base import infer_framework_load_model
|
|
|
|
| 13 |
|
| 14 |
|
| 15 |
def check_file_size(sf_filename, pt_filename):
|
|
|
|
| 17 |
pt_size = os.stat(pt_filename).st_size
|
| 18 |
|
| 19 |
if (sf_size - pt_size) / pt_size > 0.01:
|
| 20 |
+
raise RuntimeError(
|
| 21 |
+
f"""The file size different is more than 1%:
|
| 22 |
- {sf_filename}: {sf_size}
|
| 23 |
- {pt_filename}: {pt_size}
|
| 24 |
+
"""
|
| 25 |
+
)
|
| 26 |
|
| 27 |
|
| 28 |
def rename(pt_filename) -> str:
|
|
|
|
| 31 |
return local
|
| 32 |
|
| 33 |
|
| 34 |
+
def convert_multi(model_id, folder):
|
| 35 |
filename = hf_hub_download(repo_id=model_id, filename="pytorch_model.bin.index.json")
|
| 36 |
with open(filename, "r") as f:
|
| 37 |
data = json.load(f)
|
| 38 |
|
| 39 |
filenames = set(data["weight_map"].values())
|
| 40 |
+
local_filenames = []
|
| 41 |
for filename in filenames:
|
| 42 |
cached_filename = hf_hub_download(repo_id=model_id, filename=filename)
|
| 43 |
loaded = torch.load(cached_filename)
|
|
|
|
| 56 |
json.dump(newdata, f)
|
| 57 |
local_filenames.append(index)
|
| 58 |
|
| 59 |
+
operations = [
|
| 60 |
+
CommitOperationAdd(path_in_repo=local.split("/")[-1], path_or_fileobj=local) for local in local_filenames
|
| 61 |
+
]
|
| 62 |
|
| 63 |
return operations
|
| 64 |
|
|
|
|
| 76 |
operations = [CommitOperationAdd(path_in_repo=sf_filename, path_or_fileobj=local)]
|
| 77 |
return operations
|
| 78 |
|
| 79 |
+
|
| 80 |
def check_final_model(model_id, folder):
|
| 81 |
config = hf_hub_download(repo_id=model_id, filename="config.json")
|
| 82 |
shutil.copy(config, os.path.join(folder, "config.json"))
|
|
|
|
| 97 |
|
| 98 |
folder = repo_folder_name(repo_id=model_id, repo_type="models")
|
| 99 |
os.makedirs(folder)
|
| 100 |
+
new_pr = None
|
| 101 |
try:
|
| 102 |
operations = None
|
| 103 |
if "model.safetensors" in filenames or "model_index.safetensors.index.json" in filenames:
|
|
|
|
| 111 |
|
| 112 |
if operations:
|
| 113 |
check_final_model(model_id, folder)
|
| 114 |
+
new_pr = api.create_commit(
|
| 115 |
repo_id=model_id,
|
| 116 |
operations=operations,
|
| 117 |
commit_message="Adding `safetensors` variant of this model",
|
|
|
|
| 119 |
)
|
| 120 |
finally:
|
| 121 |
shutil.rmtree(folder)
|
| 122 |
+
return new_pr
|
| 123 |
|
| 124 |
|
| 125 |
if __name__ == "__main__":
|