fix: download datasets from GitHub at runtime instead of relying on package data

train.py

@@ -58,9 +58,21 @@ from deceit_env.server.grader import Grader
 from deceit_env.models import DeceitAction
 import deceit_env as _pkg
 
+# Download datasets from GitHub
+import urllib.request as _ur
+_RAW = "https://raw.githubusercontent.com/Jayant-kernel/DECEIT-the-ai-truth-environment-/main/src/deceit_env/data"
+for _fname in ["level1.jsonl", "level2.jsonl", "level3.jsonl"]:
+    _ur.urlretrieve(f"{_RAW}/{_fname}", f"/tmp/{_fname}")
+print("Datasets downloaded.")
+
 _grader = Grader(cache_path="/tmp/deceit_grader_cache.json",
                  openai_api_key=os.environ["OPENAI_API_KEY"])
-_env = DeceitEnvironment(grader=_grader)
+_env = DeceitEnvironment(
+    dataset_path="/tmp/level1.jsonl",
+    level2_dataset_path="/tmp/level2.jsonl",
+    level3_dataset_path="/tmp/level3.jsonl",
+    grader=_grader,
+)
 _env_lock = threading.Lock()
 
 # Parser
@@ -115,10 +127,8 @@ def reward_fn(completions, prompts=None, **kwargs):
     return rewards
 
 # Dataset
-import deceit_env as _de
-data_path = _pathlib.Path(_de.__file__).parent / "data" / "level1.jsonl"
 questions = []
-with open(data_path) as f:
+with open("/tmp/level1.jsonl") as f:
     for line in f:
         line = line.strip()
         if line: