File size: 1,726 Bytes
9f1a43f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
"""Auto-generated. Provides `paths` - a dict mapping each project file to a
local path your app.py can load. Files that are too big for Replit live on the
HuggingFace Hub and are downloaded on first use; everything else stays local.

Usage in app.py:

    import model_setup

    # For Keras / TensorFlow:
    import tensorflow as tf
    model = tf.keras.models.load_model(model_setup.paths["my_model.h5"])

    # For scikit-learn / joblib:
    from joblib import load
    model = load(model_setup.paths["my_model.joblib"])

    # For HuggingFace Transformers (whole directory):
    from transformers import AutoModelForSequenceClassification
    model = AutoModelForSequenceClassification.from_pretrained(model_setup.dir)

`paths` works for any filename present locally OR uploaded to HF Hub. Downloaded
Hub files are cached, so they only download once per environment.
"""
import json, os

_here = os.path.dirname(os.path.abspath(__file__))
dir = _here  # exposed for transformers/from_pretrained patterns

# 1. Every local file is reachable by its filename
paths = {fname: os.path.join(_here, fname) for fname in os.listdir(_here)}

# 2. If model_config.json exists and points to HF Hub files, download them
#    and override the local entry. If config is missing or empty, we just use
#    local files - safe fallback for projects that don't need HF Hub.
_cfg_path = os.path.join(_here, "model_config.json")
if os.path.exists(_cfg_path):
    with open(_cfg_path) as _f:
        _cfg = json.load(_f)
    if _cfg.get("hub_repo_id"):
        from huggingface_hub import hf_hub_download
        for _fname in _cfg.get("hub_files", []):
            paths[_fname] = hf_hub_download(repo_id=_cfg["hub_repo_id"], filename=_fname)