Spaces:
Running
on
CPU Upgrade
Running
on
CPU Upgrade
fixes
Browse files
app.py
CHANGED
@@ -179,13 +179,15 @@ def start_training(
|
|
179 |
else:
|
180 |
config["config"]["process"][0]["train"]["disable_sampling"] = True
|
181 |
# Save the updated config
|
182 |
-
|
|
|
|
|
183 |
with open(config_path, "w") as f:
|
184 |
yaml.dump(config, f)
|
185 |
if is_spaces:
|
186 |
print("Started training with spacerunner...")
|
187 |
-
# copy config to dataset_folder
|
188 |
-
shutil.copy(config_path, dataset_folder)
|
189 |
# get location of this script
|
190 |
script_location = os.path.dirname(os.path.abspath(__file__))
|
191 |
# copy script.py from current directory to dataset_folder
|
@@ -194,12 +196,12 @@ def start_training(
|
|
194 |
shutil.copy(script_location + "/requirements.autotrain", dataset_folder + "/requirements.txt")
|
195 |
# command to run autotrain spacerunner
|
196 |
cmd = f"autotrain spacerunner --project-name {slugged_lora_name} --script-path {dataset_folder}"
|
197 |
-
cmd += f" --username {profile.
|
198 |
outcome = subprocess.run(cmd)
|
199 |
if outcome.returncode == 0:
|
200 |
return f"""# Your training has started.
|
201 |
-
## - Training Status: <a href='https://huggingface.co/spaces/{profile.
|
202 |
-
## - Model page: <a href='https://huggingface.co/{profile.
|
203 |
else:
|
204 |
print("Error: ", outcome.stderr)
|
205 |
raise gr.Error("Something went wrong. Make sure the name of your LoRA is unique and try again")
|
|
|
179 |
else:
|
180 |
config["config"]["process"][0]["train"]["disable_sampling"] = True
|
181 |
# Save the updated config
|
182 |
+
# generate a random name for the config
|
183 |
+
random_config_name = str(uuid.uuid4())
|
184 |
+
config_path = f"/tmp/{random_config_name}-{slugged_lora_name}.yaml"
|
185 |
with open(config_path, "w") as f:
|
186 |
yaml.dump(config, f)
|
187 |
if is_spaces:
|
188 |
print("Started training with spacerunner...")
|
189 |
+
# copy config to dataset_folder as config.yaml
|
190 |
+
shutil.copy(config_path, dataset_folder + "/config.yaml")
|
191 |
# get location of this script
|
192 |
script_location = os.path.dirname(os.path.abspath(__file__))
|
193 |
# copy script.py from current directory to dataset_folder
|
|
|
196 |
shutil.copy(script_location + "/requirements.autotrain", dataset_folder + "/requirements.txt")
|
197 |
# command to run autotrain spacerunner
|
198 |
cmd = f"autotrain spacerunner --project-name {slugged_lora_name} --script-path {dataset_folder}"
|
199 |
+
cmd += f" --username {profile.username} --token {oauth_token} --backend spaces-l4x1"
|
200 |
outcome = subprocess.run(cmd)
|
201 |
if outcome.returncode == 0:
|
202 |
return f"""# Your training has started.
|
203 |
+
## - Training Status: <a href='https://huggingface.co/spaces/{profile.username}/autotrain-{slugged_lora_name}?logs=container'>{profile.username}/autotrain-{slugged_lora_name}</a> <small>(in the logs tab)</small>
|
204 |
+
## - Model page: <a href='https://huggingface.co/{profile.username}/{slugged_lora_name}'>{profile.username}/{slugged_lora_name}</a> <small>(will be available when training finishes)</small>"""
|
205 |
else:
|
206 |
print("Error: ", outcome.stderr)
|
207 |
raise gr.Error("Something went wrong. Make sure the name of your LoRA is unique and try again")
|