Spaces:
Running
Running
pedro-dev (#22)
Browse files- Fix cache removal (8d4ed6d0cc348fb02dec4066ea6aa71a5431df92)
- Set cache to a directory (383d050be2c87b73393c15023504f8e5e39e31bd)
- Convert (d5eb6e15f92173e93a9b26dbe3a935d3431c3a1c)
- Cleanup (2c3ad17d2dccee44bc09c71183dc58e1f590c06f)
- Apply quant method (dcd5ecd654cf9067d7f1efe272fe9ae6351602e5)
Co-authored-by: Pedro Cuenca <pcuenq@users.noreply.huggingface.co>
- app.py +22 -28
- cache/.keep +0 -0
- converted/.keep +0 -0
app.py
CHANGED
|
@@ -1,20 +1,17 @@
|
|
| 1 |
import os
|
| 2 |
-
import
|
| 3 |
-
|
| 4 |
-
|
| 5 |
os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
|
| 6 |
import gradio as gr
|
| 7 |
|
| 8 |
-
from huggingface_hub import
|
| 9 |
-
from huggingface_hub import snapshot_download
|
| 10 |
from huggingface_hub import whoami
|
| 11 |
from huggingface_hub import ModelCard
|
| 12 |
-
from huggingface_hub import login
|
| 13 |
from huggingface_hub import scan_cache_dir
|
| 14 |
from huggingface_hub import logging
|
| 15 |
|
| 16 |
from gradio_huggingfacehub_search import HuggingfaceHubSearch
|
| 17 |
-
|
| 18 |
from apscheduler.schedulers.background import BackgroundScheduler
|
| 19 |
|
| 20 |
from textwrap import dedent
|
|
@@ -22,23 +19,24 @@ from textwrap import dedent
|
|
| 22 |
import mlx_lm
|
| 23 |
from mlx_lm import convert
|
| 24 |
|
| 25 |
-
from typing import Any, Callable, Dict, Generator, List, Optional, Tuple, Type, Union
|
| 26 |
-
|
| 27 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
| 28 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 29 |
def clear_hf_cache_space():
|
| 30 |
scan = scan_cache_dir()
|
| 31 |
to_delete = []
|
| 32 |
for repo in scan.repos:
|
| 33 |
if repo.repo_type == "model":
|
| 34 |
-
to_delete.
|
| 35 |
-
|
| 36 |
-
scan.delete_revisions(to_delete)
|
| 37 |
-
|
| 38 |
print("Cache has been cleared")
|
| 39 |
|
| 40 |
def upload_to_hub(path, upload_repo, hf_path, token):
|
| 41 |
-
|
| 42 |
card = ModelCard.load(hf_path)
|
| 43 |
card.data.tags = ["mlx"] if card.data.tags is None else card.data.tags + ["mlx"]
|
| 44 |
card.data.base_model = hf_path
|
|
@@ -86,33 +84,29 @@ def upload_to_hub(path, upload_repo, hf_path, token):
|
|
| 86 |
)
|
| 87 |
print(f"Upload successful, go to https://huggingface.co/{upload_repo} for details.")
|
| 88 |
|
| 89 |
-
def process_model(model_id, q_method,oauth_token: gr.OAuthToken | None):
|
| 90 |
-
|
| 91 |
if oauth_token.token is None:
|
| 92 |
raise ValueError("You must be logged in to use MLX-my-repo")
|
| 93 |
|
| 94 |
model_name = model_id.split('/')[-1]
|
| 95 |
-
print(model_name)
|
| 96 |
username = whoami(oauth_token.token)["name"]
|
| 97 |
-
print(username)
|
| 98 |
-
|
| 99 |
-
# login(token=oauth_token.token, add_to_git_credential=True)
|
| 100 |
-
|
| 101 |
try:
|
| 102 |
-
upload_repo = username
|
| 103 |
print(upload_repo)
|
| 104 |
-
|
| 105 |
-
|
| 106 |
-
|
| 107 |
-
|
|
|
|
|
|
|
|
|
|
| 108 |
return (
|
| 109 |
-
f'Find your repo <a href
|
| 110 |
"llama.png",
|
| 111 |
)
|
| 112 |
except Exception as e:
|
| 113 |
return (f"Error: {e}", "error.png")
|
| 114 |
finally:
|
| 115 |
-
shutil.rmtree("mlx_model", ignore_errors=True)
|
| 116 |
clear_hf_cache_space()
|
| 117 |
print("Folder cleaned up successfully!")
|
| 118 |
|
|
|
|
| 1 |
import os
|
| 2 |
+
import tempfile
|
| 3 |
+
|
| 4 |
+
os.environ["HF_HUB_CACHE"] = "cache"
|
| 5 |
os.environ["GRADIO_ANALYTICS_ENABLED"] = "False"
|
| 6 |
import gradio as gr
|
| 7 |
|
| 8 |
+
from huggingface_hub import HfApi
|
|
|
|
| 9 |
from huggingface_hub import whoami
|
| 10 |
from huggingface_hub import ModelCard
|
|
|
|
| 11 |
from huggingface_hub import scan_cache_dir
|
| 12 |
from huggingface_hub import logging
|
| 13 |
|
| 14 |
from gradio_huggingfacehub_search import HuggingfaceHubSearch
|
|
|
|
| 15 |
from apscheduler.schedulers.background import BackgroundScheduler
|
| 16 |
|
| 17 |
from textwrap import dedent
|
|
|
|
| 19 |
import mlx_lm
|
| 20 |
from mlx_lm import convert
|
| 21 |
|
|
|
|
|
|
|
| 22 |
HF_TOKEN = os.environ.get("HF_TOKEN")
|
| 23 |
|
| 24 |
+
# I'm not sure if we need to add more stuff here
|
| 25 |
+
QUANT_PARAMS = {
|
| 26 |
+
"Q4": 4,
|
| 27 |
+
"Q8": 8,
|
| 28 |
+
}
|
| 29 |
+
|
| 30 |
def clear_hf_cache_space():
|
| 31 |
scan = scan_cache_dir()
|
| 32 |
to_delete = []
|
| 33 |
for repo in scan.repos:
|
| 34 |
if repo.repo_type == "model":
|
| 35 |
+
to_delete.extend([rev.commit_hash for rev in repo.revisions])
|
| 36 |
+
scan.delete_revisions(*to_delete).execute()
|
|
|
|
|
|
|
| 37 |
print("Cache has been cleared")
|
| 38 |
|
| 39 |
def upload_to_hub(path, upload_repo, hf_path, token):
|
|
|
|
| 40 |
card = ModelCard.load(hf_path)
|
| 41 |
card.data.tags = ["mlx"] if card.data.tags is None else card.data.tags + ["mlx"]
|
| 42 |
card.data.base_model = hf_path
|
|
|
|
| 84 |
)
|
| 85 |
print(f"Upload successful, go to https://huggingface.co/{upload_repo} for details.")
|
| 86 |
|
| 87 |
+
def process_model(model_id, q_method, oauth_token: gr.OAuthToken | None):
|
|
|
|
| 88 |
if oauth_token.token is None:
|
| 89 |
raise ValueError("You must be logged in to use MLX-my-repo")
|
| 90 |
|
| 91 |
model_name = model_id.split('/')[-1]
|
|
|
|
| 92 |
username = whoami(oauth_token.token)["name"]
|
|
|
|
|
|
|
|
|
|
|
|
|
| 93 |
try:
|
| 94 |
+
upload_repo = f"{username}/{model_name}-{q_method}-mlx"
|
| 95 |
print(upload_repo)
|
| 96 |
+
with tempfile.TemporaryDirectory(dir="converted") as tmpdir:
|
| 97 |
+
# The target dir must not exist
|
| 98 |
+
mlx_path = os.path.join(tmpdir, "mlx")
|
| 99 |
+
convert(model_id, mlx_path=mlx_path, quantize=True, q_bits=QUANT_PARAMS[q_method])
|
| 100 |
+
print("Conversion done")
|
| 101 |
+
upload_to_hub(path=mlx_path, upload_repo=upload_repo, hf_path=model_id, token=oauth_token.token)
|
| 102 |
+
print("Upload done")
|
| 103 |
return (
|
| 104 |
+
f'Find your repo <a href="https://hf.co/{upload_repo}" target="_blank" style="text-decoration:underline">here</a>',
|
| 105 |
"llama.png",
|
| 106 |
)
|
| 107 |
except Exception as e:
|
| 108 |
return (f"Error: {e}", "error.png")
|
| 109 |
finally:
|
|
|
|
| 110 |
clear_hf_cache_space()
|
| 111 |
print("Folder cleaned up successfully!")
|
| 112 |
|
cache/.keep
ADDED
|
File without changes
|
converted/.keep
ADDED
|
File without changes
|