Upload push_code_to_hf.py with huggingface_hub

push_code_to_hf.py

@@ -11,6 +11,7 @@ Usage:
 
 import argparse
 import os
+from pathlib import Path
 from huggingface_hub import HfApi, create_repo
 
 DEFAULT_CODE_REPO = "ASTERIZER/LUNA"
@@ -20,6 +21,7 @@ TOKEN = os.environ.get("HF_TOKEN")
 FILES_TO_PUSH = [
     # Core training scripts
     "train.py",
+    "train_300m.py",
     "sft_train.py",
     "lora_sft_train.py",
     "chat.py",
@@ -28,6 +30,7 @@ FILES_TO_PUSH = [
 
     # Configs
     "train_config.yaml",
+    "train_config_300m.yaml",
     "train_continue_english_1b.yaml",
     "sft_config.yaml",
     "rag_mcp_lora_config.yaml",
@@ -59,6 +62,7 @@ FILES_TO_PUSH = [
     # Instance run scripts
     "run_english_1b_instance.sh",
     "setup_and_train.sh",
+    "setup_and_train_300m.sh",
     "setup_and_sft.sh",
     "gpu_train.sh",
     "gpu_full_sft.sh",
@@ -83,6 +87,37 @@ FILES_TO_PUSH = [
     "Base/checkpoints/EleutherAI/pythia-160m/tokenizer.json",
 ]
 
+EVALUATION_GLOBS = [
+    "Evaluation/*.py",
+    "Evaluation/*.sh",
+    "Evaluation/*.yaml",
+    "Evaluation/*.yml",
+    "Evaluation/*.md",
+    "Evaluation/*.txt",
+    "Evaluation/*.json",
+]
+
+
+def build_file_list():
+    files = []
+    seen = set()
+
+    for fpath in FILES_TO_PUSH:
+        if fpath not in seen:
+            files.append(fpath)
+            seen.add(fpath)
+
+    for pattern in EVALUATION_GLOBS:
+        for path in sorted(Path(".").glob(pattern)):
+            if not path.is_file():
+                continue
+            rel = path.as_posix()
+            if rel not in seen:
+                files.append(rel)
+                seen.add(rel)
+
+    return files
+
 
 def main():
     parser = argparse.ArgumentParser(description="Push LUNA code to HuggingFace")
@@ -107,8 +142,11 @@ def main():
     )
     print(f"Repo ready: https://huggingface.co/{'spaces/' if args.type == 'space' else ''}{args.repo}")
 
+    files_to_push = build_file_list()
+    print(f"Preparing to push {len(files_to_push)} files...")
+
     pushed = 0
-    for fpath in FILES_TO_PUSH:
+    for fpath in files_to_push:
         if not os.path.exists(fpath):
             print(f"  SKIP (not found): {fpath}")
             continue
@@ -122,7 +160,7 @@ def main():
         print(f"  OK: {fpath}")
         pushed += 1
 
-    print(f"\nPushed {pushed}/{len(FILES_TO_PUSH)} files to https://huggingface.co/{'spaces/' if args.type == 'space' else ''}{args.repo}")
+    print(f"\nPushed {pushed}/{len(files_to_push)} files to https://huggingface.co/{'spaces/' if args.type == 'space' else ''}{args.repo}")
 
 
 if __name__ == "__main__":