Spaces:
Running
on
A10G
Running
on
A10G
Remove references to convert.py (#73)
Browse files- Remove references to convert.py (cb0a2b8372fbdccbfc50f8001503a36898f34738)
app.py
CHANGED
@@ -15,19 +15,8 @@ from apscheduler.schedulers.background import BackgroundScheduler
|
|
15 |
|
16 |
from textwrap import dedent
|
17 |
|
18 |
-
LLAMA_LIKE_ARCHS = ["MistralForCausalLM",]
|
19 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
20 |
|
21 |
-
def script_to_use(model_id, api):
|
22 |
-
info = api.model_info(model_id)
|
23 |
-
if info.config is None:
|
24 |
-
return None
|
25 |
-
arch = info.config.get("architectures", None)
|
26 |
-
if arch is None:
|
27 |
-
return None
|
28 |
-
arch = arch[0]
|
29 |
-
return "convert.py" if arch in LLAMA_LIKE_ARCHS else "convert-hf-to-gguf.py"
|
30 |
-
|
31 |
def split_upload_model(model_path, repo_id, oauth_token: gr.OAuthToken | None, split_max_tensors=256, split_max_size=None):
|
32 |
if oauth_token.token is None:
|
33 |
raise ValueError("You have to be logged in.")
|
@@ -98,7 +87,7 @@ def process_model(model_id, q_method, private_repo, split_model, split_max_tenso
|
|
98 |
print(f"Current working directory: {os.getcwd()}")
|
99 |
print(f"Model directory contents: {os.listdir(model_name)}")
|
100 |
|
101 |
-
conversion_script =
|
102 |
fp16_conversion = f"python llama.cpp/{conversion_script} {model_name} --outtype f16 --outfile {fp16}"
|
103 |
result = subprocess.run(fp16_conversion, shell=True, capture_output=True)
|
104 |
print(result)
|
|
|
15 |
|
16 |
from textwrap import dedent
|
17 |
|
|
|
18 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
19 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
20 |
def split_upload_model(model_path, repo_id, oauth_token: gr.OAuthToken | None, split_max_tensors=256, split_max_size=None):
|
21 |
if oauth_token.token is None:
|
22 |
raise ValueError("You have to be logged in.")
|
|
|
87 |
print(f"Current working directory: {os.getcwd()}")
|
88 |
print(f"Model directory contents: {os.listdir(model_name)}")
|
89 |
|
90 |
+
conversion_script = "convert-hf-to-gguf.py"
|
91 |
fp16_conversion = f"python llama.cpp/{conversion_script} {model_name} --outtype f16 --outfile {fp16}"
|
92 |
result = subprocess.run(fp16_conversion, shell=True, capture_output=True)
|
93 |
print(result)
|