structverify 0.3.1__tar.gz → 0.3.2__tar.gz

{structverify-0.3.1/structverify.egg-info → structverify-0.3.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: structverify
-Version: 0.3.1
+Version: 0.3.2
 Summary: Compliance & fact verification for documents — check your files against your own rulebook (PDF) or a data source, and read a plain True/False.
 Author-email: "김예슬 (Yeseul Kim)" <yesul0718@gmail.com>
 License: MIT License

{structverify-0.3.1 → structverify-0.3.2}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "structverify"
-version = "0.3.1"
+version = "0.3.2"
 description = "Compliance & fact verification for documents — check your files against your own rulebook (PDF) or a data source, and read a plain True/False."
 readme = "README.md"
 license = { file = "LICENSE" }

{structverify-0.3.1 → structverify-0.3.2}/structverify/__init__.py

@@ -54,7 +54,7 @@ _TRAINING_SYMBOLS = frozenset({
 })
 
 __all__ = sorted(_API_SYMBOLS | _PIPELINE_SYMBOLS | _PROGRESS_SYMBOLS | _TRAINING_SYMBOLS)
-__version__ = "0.3.1"
+__version__ = "0.3.2"
 
 
 def __getattr__(name: str):

{structverify-0.3.1 → structverify-0.3.2}/structverify/training/monitor.py

@@ -10,11 +10,34 @@
 """
 from __future__ import annotations
 
+import os
+import sys
 import time
 from dataclasses import dataclass, field
 
 _SPARK = "▁▂▃▄▅▆▇█"
 
+# ── 터미널 색 (v0.3.2 팔레트) ────────────────────────────────────────────────
+_ANSI = {"cyan": "\x1b[38;5;117m", "spark": "\x1b[38;5;111m", "dim": "\x1b[2m",
+         "bold": "\x1b[1m", "green": "\x1b[38;5;114m", "amber": "\x1b[38;5;179m",
+         "red": "\x1b[38;5;210m", "reset": "\x1b[0m"}
+
+
+def _color_enabled() -> bool:
+    """NO_COLOR면 끔, SV_COLOR=1이면 강제 켬, 그 외엔 tty 감지."""
+    if os.environ.get("NO_COLOR"):
+        return False
+    if os.environ.get("SV_COLOR"):
+        return True
+    try:
+        return sys.stderr.isatty()
+    except Exception:  # noqa: BLE001
+        return False
+
+
+def _c(name: str, s: str, on: bool) -> str:
+    return f"{_ANSI[name]}{s}{_ANSI['reset']}" if on else s
+
 
 def sparkline(values: list[float], width: int = 8) -> str:
     """최근 값들을 ▁▂▃ 스파크라인으로. 값이 없으면 빈 문자열."""
@@ -118,19 +141,29 @@ class LossMonitor:
 
     # ── 터미널 UI ───────────────────────────────────────────────────────────
     def render(self, max_steps: int, started_at: float, lr: float | None = None,
-               width: int = 12) -> str:
-        """현재 상태를 터미널 한 줄로."""
+               width: int = 12, color: bool | None = None) -> str:
+        """현재 상태를 터미널 한 줄로 (tty면 색 포함 — NO_COLOR/SV_COLOR로 제어)."""
+        on = _color_enabled() if color is None else color
         frac = self.step / max_steps if max_steps else 0.0
         filled = int(frac * width)
         bar = "▓" * filled + "░" * (width - filled)
         loss = self.losses[-1] if self.losses else float("nan")
         spark = sparkline(self.means)
         elapsed = time.time() - started_at
-        eta = _fmt_eta(elapsed / frac - elapsed) if frac > 0.01 else "--:--"
-        lr_s = f" · lr {lr:.1e}" if lr else ""
-        spike_s = f" · ⚡{len(self.spikes)}" if self.spikes else ""
-        return (f"🧠 {self.step}/{max_steps} {bar} {frac*100:3.0f}% · "
-                f"loss {loss:.3f} {spark}{lr_s} · ETA {eta}{spike_s} · 🩺 {self.status}")
+        eta = _fmt_eta(elapsed / frac - elapsed) if frac > 0.005 else "--:--"
+        st_color = ("red" if "발산" in self.status
+                    else "amber" if self.status == "워밍업" else "green")
+        parts = [
+            "🧠 " + _c("cyan", f"{self.step}/{max_steps} {bar} {frac*100:3.0f}%", on),
+            _c("bold", f"loss {loss:.3f}", on) + " " + _c("spark", spark, on),
+        ]
+        if lr:
+            parts.append(_c("dim", f"lr {lr:.1e}", on))
+        parts.append(_c("dim", f"ETA {eta}", on))
+        if self.spikes:
+            parts.append(_c("amber", f"⚡{len(self.spikes)}", on))
+        parts.append("🩺 " + _c(st_color, self.status, on))
+        return " · ".join(parts)
 
     def summary(self) -> str:
         """학습 종료 시 요약 (레시피가 마지막에 출력)."""

{structverify-0.3.1 → structverify-0.3.2}/structverify/training/recipe/train_qlora.py

@@ -26,6 +26,17 @@ os.environ.setdefault("TF_CPP_MIN_LOG_LEVEL", "3")
 os.environ.setdefault("TRANSFORMERS_VERBOSITY", "error")
 os.environ.setdefault("TOKENIZERS_PARALLELISM", "false")
 
+# "Exception ignored in: <_io.BytesIO ...> BufferError" 도배 차단.
+#   Python 3.13 + datasets 멀티프로세스 토큰화의 무해한 정리 단계 잡음인데,
+#   워커가 fork로 뜨면서 이 훅을 물려받아 자식 프로세스 출력까지 조용해진다.
+sys.unraisablehook = lambda *_: None
+
+import logging
+import warnings
+
+warnings.filterwarnings("ignore")                       # Flax deprecated 등
+logging.getLogger("torchao").setLevel(logging.CRITICAL)  # torchao .so 로드 WARNING
+
 
 def load_chat_dataset(path):
     from datasets import Dataset
@@ -65,6 +76,16 @@ def main():
     from unsloth import FastLanguageModel
     from trl import SFTTrainer, SFTConfig
 
+    # tqdm 진행바(체크포인트 로드·Map·Tokenizing) 대신 우리 phase 라인으로.
+    try:
+        import datasets as _ds
+        import transformers as _tf
+        _ds.disable_progress_bars()
+        _tf.utils.logging.disable_progress_bar()
+    except Exception:  # noqa: BLE001
+        pass
+
+    print(f"⚙ 베이스 모델 로드 — {args.model} (4bit)", file=sys.stderr)
     model, tokenizer = FastLanguageModel.from_pretrained(
         model_name=args.model,
         max_seq_length=args.seq_len,
@@ -81,6 +102,7 @@ def main():
         use_gradient_checkpointing="unsloth",
     )
 
+    print("⚙ 데이터 준비·토큰화 중 …", file=sys.stderr)
     ds = load_chat_dataset(args.data)
 
     def fmt(ex):
@@ -112,7 +134,7 @@ def main():
     from transformers import TrainerCallback
     from transformers.trainer_callback import PrinterCallback
 
-    from structverify.training.monitor import LossMonitor
+    from structverify.training.monitor import LossMonitor, _c, _color_enabled
 
     class SVLiveCallback(TrainerCallback):
         def __init__(self, early_stop: bool, patience: int):
@@ -128,18 +150,20 @@ def main():
             print("\r\x1b[2K" + line, end="", flush=True, file=sys.stderr)
             if self.mon.should_stop and self.early_stop:
                 print("", file=sys.stderr)
-                print(f"✂ {self.mon.stop_reason}", file=sys.stderr)
+                print(_c("amber", f"✂ {self.mon.stop_reason}", _color_enabled()), file=sys.stderr)
                 control.should_training_stop = True
             return control
 
         def on_train_end(self, targs, state, control, **kw):
             print("", file=sys.stderr)
-            print(self.mon.summary(), file=sys.stderr)
+            print(_c("green", self.mon.summary(), _color_enabled()), file=sys.stderr)
             return control
 
     trainer.remove_callback(PrinterCallback)     # {'loss': ...} dict 도배 제거
     trainer.add_callback(SVLiveCallback(args.early_stop, args.patience))
 
+    es = "ON" if args.early_stop else "OFF (관찰만)"
+    print(f"🩺 실시간 감독 시작 — patience {args.patience} · 조기 종료 {es}", file=sys.stderr)
     trainer.train()
 
     # 어댑터 + trainer_state.json 저장 (TrainDoctor가 읽음)

{structverify-0.3.1 → structverify-0.3.2/structverify.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: structverify
-Version: 0.3.1
+Version: 0.3.2
 Summary: Compliance & fact verification for documents — check your files against your own rulebook (PDF) or a data source, and read a plain True/False.
 Author-email: "김예슬 (Yeseul Kim)" <yesul0718@gmail.com>
 License: MIT License