Spaces:
Runtime error
Runtime error
better error message + fail early
Browse files- app.py +39 -43
- pyproject.toml +2 -0
app.py
CHANGED
|
@@ -1,4 +1,6 @@
|
|
| 1 |
import pathlib
|
|
|
|
|
|
|
| 2 |
import tempfile
|
| 3 |
from typing import Iterable, List
|
| 4 |
|
|
@@ -36,9 +38,7 @@ has_gpu = torch.cuda.is_available()
|
|
| 36 |
# )
|
| 37 |
|
| 38 |
cli = "mergekit-yaml config.yaml merge --copy-tokenizer" + (
|
| 39 |
-
" --cuda --low-cpu-memory"
|
| 40 |
-
if has_gpu
|
| 41 |
-
else " --allow-crimes --out-shard-size 1B --lazy-unpickle"
|
| 42 |
)
|
| 43 |
|
| 44 |
## This Space is heavily inspired by LazyMergeKit by Maxime Labonne
|
|
@@ -89,17 +89,22 @@ A quick overview of the currently supported merge methods:
|
|
| 89 |
examples = [[str(f)] for f in pathlib.Path("examples").glob("*.yml")]
|
| 90 |
|
| 91 |
|
| 92 |
-
def merge(
|
| 93 |
-
|
| 94 |
-
|
| 95 |
if not yaml_config:
|
| 96 |
-
|
|
|
|
| 97 |
try:
|
| 98 |
merge_config = MergeConfiguration.model_validate(yaml.safe_load(yaml_config))
|
| 99 |
except Exception as e:
|
| 100 |
-
|
|
|
|
| 101 |
|
| 102 |
-
|
|
|
|
|
|
|
|
|
|
| 103 |
|
| 104 |
with tempfile.TemporaryDirectory() as tmpdirname:
|
| 105 |
tmpdir = pathlib.Path(tmpdirname)
|
|
@@ -109,44 +114,35 @@ def merge(
|
|
| 109 |
config_path.write_text(yaml_config)
|
| 110 |
yield runner.log(f"Merge configuration saved in {config_path}")
|
| 111 |
|
| 112 |
-
if
|
| 113 |
-
name
|
| 114 |
-
|
| 115 |
-
)
|
| 116 |
-
repo_name
|
| 117 |
-
|
| 118 |
-
|
| 119 |
-
|
| 120 |
-
|
| 121 |
-
|
| 122 |
-
|
| 123 |
-
|
| 124 |
-
yield runner.log(
|
| 125 |
-
|
| 126 |
-
)
|
| 127 |
|
| 128 |
yield from runner.run_command(cli.split(), cwd=merged_path)
|
| 129 |
|
| 130 |
if runner.exit_code != 0:
|
| 131 |
-
yield runner.log(
|
| 132 |
-
|
| 133 |
-
)
|
| 134 |
return
|
| 135 |
|
| 136 |
-
|
| 137 |
-
|
| 138 |
-
|
| 139 |
-
|
| 140 |
-
|
| 141 |
-
|
| 142 |
-
|
| 143 |
-
print("Starting upload")
|
| 144 |
-
folder_url = api.upload_folder(
|
| 145 |
-
repo_id=repo_url.repo_id, folder_path=merged_path / "merge"
|
| 146 |
-
)
|
| 147 |
-
print(f"Model successfully uploaded to {folder_url}")
|
| 148 |
-
|
| 149 |
-
yield from runner.run_python(upload_to_repo)
|
| 150 |
|
| 151 |
|
| 152 |
with gr.Blocks() as demo:
|
|
@@ -161,12 +157,12 @@ with gr.Blocks() as demo:
|
|
| 161 |
label="HF Write Token",
|
| 162 |
info="https://hf.co/settings/token",
|
| 163 |
type="password",
|
| 164 |
-
placeholder="
|
| 165 |
)
|
| 166 |
repo_name = gr.Textbox(
|
| 167 |
lines=1,
|
| 168 |
label="Repo name",
|
| 169 |
-
placeholder="
|
| 170 |
)
|
| 171 |
button = gr.Button("Merge", variant="primary")
|
| 172 |
logs = LogsView()
|
|
|
|
| 1 |
import pathlib
|
| 2 |
+
import random
|
| 3 |
+
import string
|
| 4 |
import tempfile
|
| 5 |
from typing import Iterable, List
|
| 6 |
|
|
|
|
| 38 |
# )
|
| 39 |
|
| 40 |
cli = "mergekit-yaml config.yaml merge --copy-tokenizer" + (
|
| 41 |
+
" --cuda --low-cpu-memory" if has_gpu else " --allow-crimes --out-shard-size 1B --lazy-unpickle"
|
|
|
|
|
|
|
| 42 |
)
|
| 43 |
|
| 44 |
## This Space is heavily inspired by LazyMergeKit by Maxime Labonne
|
|
|
|
| 89 |
examples = [[str(f)] for f in pathlib.Path("examples").glob("*.yml")]
|
| 90 |
|
| 91 |
|
| 92 |
+
def merge(yaml_config: str, hf_token: str, repo_name: str) -> Iterable[List[Log]]:
|
| 93 |
+
runner = LogsViewRunner()
|
| 94 |
+
|
| 95 |
if not yaml_config:
|
| 96 |
+
yield runner.log("Empty yaml, pick an example below", level="ERROR")
|
| 97 |
+
return
|
| 98 |
try:
|
| 99 |
merge_config = MergeConfiguration.model_validate(yaml.safe_load(yaml_config))
|
| 100 |
except Exception as e:
|
| 101 |
+
yield runner.log(f"Invalid yaml {e}", level="ERROR")
|
| 102 |
+
return
|
| 103 |
|
| 104 |
+
if not hf_token:
|
| 105 |
+
yield runner.log("You must provide a write-access token.", level="ERROR")
|
| 106 |
+
return
|
| 107 |
+
api = huggingface_hub.HfApi(token=hf_token)
|
| 108 |
|
| 109 |
with tempfile.TemporaryDirectory() as tmpdirname:
|
| 110 |
tmpdir = pathlib.Path(tmpdirname)
|
|
|
|
| 114 |
config_path.write_text(yaml_config)
|
| 115 |
yield runner.log(f"Merge configuration saved in {config_path}")
|
| 116 |
|
| 117 |
+
if not repo_name:
|
| 118 |
+
yield runner.log("No repo name provided. Generating a random one.")
|
| 119 |
+
repo_name = f"mergekit-{merge_config.merge_method}"
|
| 120 |
+
# Make repo_name "unique" (no need to be extra careful on uniqueness)
|
| 121 |
+
repo_name += "-" + "".join(random.choices(string.ascii_lowercase, k=7))
|
| 122 |
+
repo_name = repo_name.replace("/", "-").strip("-")
|
| 123 |
+
|
| 124 |
+
try:
|
| 125 |
+
yield runner.log(f"Creating repo {repo_name}")
|
| 126 |
+
repo_url = api.create_repo(repo_name, exist_ok=True)
|
| 127 |
+
yield runner.log(f"Repo created: {repo_url}")
|
| 128 |
+
except Exception as e:
|
| 129 |
+
yield runner.log(f"Error creating repo {e}", level="ERROR")
|
| 130 |
+
return
|
|
|
|
| 131 |
|
| 132 |
yield from runner.run_command(cli.split(), cwd=merged_path)
|
| 133 |
|
| 134 |
if runner.exit_code != 0:
|
| 135 |
+
yield runner.log("Merge failed. Deleting repo as no model is uploaded.", level="ERROR")
|
| 136 |
+
api.delete_repo(repo_url.repo_id)
|
|
|
|
| 137 |
return
|
| 138 |
|
| 139 |
+
yield runner.log("Model merged successfully. Uploading to HF.")
|
| 140 |
+
yield from runner.run_python(
|
| 141 |
+
api.upload_folder,
|
| 142 |
+
repo_id=repo_url.repo_id,
|
| 143 |
+
folder_path=merged_path / "merge",
|
| 144 |
+
)
|
| 145 |
+
yield runner.log("Model successfully uploaded to HF.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 146 |
|
| 147 |
|
| 148 |
with gr.Blocks() as demo:
|
|
|
|
| 157 |
label="HF Write Token",
|
| 158 |
info="https://hf.co/settings/token",
|
| 159 |
type="password",
|
| 160 |
+
placeholder="Mandatory. Used to upload the merged model.",
|
| 161 |
)
|
| 162 |
repo_name = gr.Textbox(
|
| 163 |
lines=1,
|
| 164 |
label="Repo name",
|
| 165 |
+
placeholder="Optional. Will create a random name if empty.",
|
| 166 |
)
|
| 167 |
button = gr.Button("Merge", variant="primary")
|
| 168 |
logs = LogsView()
|
pyproject.toml
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
[tool.ruff]
|
| 2 |
+
line-length=119
|