Upload model support files
Browse files
config.py
ADDED
|
@@ -0,0 +1,34 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
+
import torch
|
| 4 |
+
|
| 5 |
+
DATA_DIR = "data"
|
| 6 |
+
TOKENIZER_PATH = "tokenizer.json"
|
| 7 |
+
CHECKPOINT_DIR = "checkpoints"
|
| 8 |
+
TRAIN_TOKENS_CACHE = "data/train_tokens.npy"
|
| 9 |
+
VAL_TOKENS_CACHE = "data/val_tokens.npy"
|
| 10 |
+
|
| 11 |
+
DEVICE = "cuda" if torch.cuda.is_available() else "cpu"
|
| 12 |
+
DTYPE = torch.float16 if DEVICE == "cuda" else torch.float32
|
| 13 |
+
|
| 14 |
+
|
| 15 |
+
def load_dotenv(path=".env"):
|
| 16 |
+
"""Tiny dependency-free .env loader. Populates os.environ (does not overwrite
|
| 17 |
+
variables already set in the real environment). Lets HF_TOKEN live in a
|
| 18 |
+
gitignored .env file instead of being hardcoded or pasted into the shell."""
|
| 19 |
+
if not os.path.exists(path):
|
| 20 |
+
return
|
| 21 |
+
with open(path, encoding="utf-8") as f:
|
| 22 |
+
for line in f:
|
| 23 |
+
line = line.strip()
|
| 24 |
+
if not line or line.startswith("#") or "=" not in line:
|
| 25 |
+
continue
|
| 26 |
+
key, _, value = line.partition("=")
|
| 27 |
+
key, value = key.strip(), value.strip().strip('"').strip("'")
|
| 28 |
+
os.environ.setdefault(key, value)
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
load_dotenv()
|
| 32 |
+
|
| 33 |
+
# HuggingFace token (optional for public datasets like wikitext-103; used if present).
|
| 34 |
+
HF_TOKEN = os.environ.get("HF_TOKEN") or os.environ.get("HUGGINGFACE_TOKEN")
|