Autoresearch 특징 발견
TypeSafe 질문을 제안하고, 자유 텍스트를 수치적 특징으로 변환하며, 모델 오류를 이용해 지도 학습된 CatBoost 회귀 모델을 개선하는 autoresearch 루프를 실행합니다.
TypeSafe 질문은 지도 학습된 CatBoost 모델을 위해 자유 텍스트를 수치적 특징으로 바꿉니다. autoresearch 루프로 그 질문들을 발견합니다.
CatBoost에는 숫자 표가 필요하지만, 시음 노트는 그런 표가 아닙니다. 이 cookbook은 노트에 관한 질문들로 그 표를 만듭니다. 그중 어느 질문도 손으로 쓰지 않습니다. LLM이 질문을 제안하고, TypeSafe가 모든 행에 대해 답하며, CatBoost가 그 답으로 학습합니다. autoresearch 부분은 그다음입니다. CatBoost가 어떤 질문을 썼고 어떤 행을 여전히 틀리는지 보고하면, 다음 제안 호출이 그 보고를 읽고, 루프가 다시 돕니다.
끝나면 자신의 레이블된 텍스트에 겨눌 수 있는 루프, 라운드별 홀드아웃 오류 곡선, 그리고 최종 모델이 가장 많이 쓴 질문 표를 얻게 됩니다.
tasting note
|
v
38 TypeSafe answers
|-- 29 score questions x 2 columns = 58
| expected rubric level + answer uncertainty
`-- 9 noul questions x 1 column = 9
probability true
|
v
67 numeric columns --> CatBoost --> predicted critic score
held-out RMSE: 1.77 pointsscore 답은 두 개의 열이 됩니다. 답이 가리키는 평균 레벨과, 그 평균 주위에 얼마나 퍼져 있는지입니다. noul 답은 확률 하나이므로 열 하나입니다.
데이터는 와인 리뷰 2,000건입니다. 시음 노트가 입력이고, 80-100 척도의 평론가 점수가 출력입니다. RMSE는 예측 오류를 평론가 점수(점) 단위로 측정하며, 크게 빗나갈수록 더 무겁게 세고, 낮을수록 좋습니다. 아래 표의 모든 수치는 모델도 루프도 한 번도 본 적 없는 리뷰 800건에서 나온 것입니다.
| 노트가 점수가 되는 방식 | RMSE |
|---|---|
| 훈련 행의 평균 점수를 예측 | 3.09 |
| 같은 CatBoost가 노트를 단어 수로 읽음 | 2.47 |
| TypeSafe에 점수 자체를 요청, 재조정하고 이동 | 2.15 |
| 제안 호출 한 번에서 나온 질문 18개, 루프 없음 | 1.87 |
| 루프 5라운드 후의 질문 38개 | 1.77 |
마지막 두 행이 루프입니다. 아직 아무 근거도 없는 제안 호출 한 번이 1.87까지 갑니다. 자기 최악 예측을 읽는 네 라운드가 더해져 1.77까지 갑니다. 이득의 대부분은 그 첫 호출에 있고, 그 뒤 네 라운드가 얼마를 더하는지는 아래에서 측정합니다.
from __future__ import annotations
import json
import os
import random
import textwrap
import urllib.request
from concurrent.futures import ThreadPoolExecutor
from pathlib import Path
from time import perf_counter
from typing import NamedTuple
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
from catboost import CatBoostRegressor
from cooksafe import JsonCache, make_playground_link
from IPython.display import Markdown, display
from typesafe_sdk import Noul, NoulCriteria, Score, TypeSafeClient
matplotlib.use("Agg") # headless render
TYPESAFE_MODEL = "jev-1.12"
FOLDS, REPEATS = 5, 3 # repeats steady the error at this sample size
CATBOOST = dict(
iterations=400,
depth=4,
learning_rate=0.05,
loss_function="RMSE",
verbose=0,
random_seed=0,
thread_count=1,
allow_writing_files=False,
)
client = TypeSafeClient(
# keyless kernels replay the cache
api_key=os.environ.get("TYPESAFE_API_KEY", "cache-only"),
base_url=os.environ.get("TYPESAFE_ENDPOINT"),
timeout=120.0,
)
json_cache = JsonCache(Path("json_cache.json"))
# ----------------------------------------------------------------- the specification
INTENSITY_LEVELS = [
"Not present in this note at all",
"Barely present - mentioned once, in passing",
"Present at a moderate level",
"Present strongly - the note dwells on it",
"Dominant - the note is largely about this",
]
PRESENCE_CRITERIA = NoulCriteria(
true="The note states this or clearly implies it",
false="The note gives no indication of this",
)
# Asking for the score outright: ten quality bands, rescaled onto the 80-100 critic scale.
SCORE_LEVELS = [
"Faulty or unpleasant - the note is mostly criticism",
"Barely acceptable - drinkable, with nothing to recommend it",
"Simple and sound - correct, plain, forgettable",
"Pleasant everyday wine - some appeal, little depth",
"Good - clear varietal character, well made",
"Very good - balanced, with something to say",
"Excellent - complex and structured",
"Outstanding - depth and length, built to age",
"Superb - among the best of its type",
"Profound - the note treats it as exceptional",
]
# Structured output requires every property in `required`, so unused fields come back empty.
PROPOSAL_SCHEMA = {
"type": "object",
"properties": {
"actions": {
"type": "array",
"items": {
"type": "object",
"properties": {
"op": {"type": "string", "enum": ["add", "revise", "drop"]},
"target": {"type": "string"},
"name": {"type": "string"},
"kind": {"type": "string", "enum": ["intensity", "presence"]},
"question": {"type": "string"},
},
"required": ["op", "target", "name", "kind", "question"],
"additionalProperties": False,
},
}
},
"required": ["actions"],
"additionalProperties": False,
}
PROPOSALS = 18 # actions the proposer may return per round
# The one string that knows this is about wine. Point it at your own label and text.
PROPOSER_TASK = f"""You are designing numeric features for a gradient-boosting model that
predicts the score a wine critic gave (an integer from 80 to 100) from the tasting note alone.
The model sees nothing but the features you design.
Return up to {PROPOSALS} actions. Each action is one of:
- {{"op": "add", "target": "", "name": ..., "kind": ..., "question": ...}}
A new feature.
- {{"op": "revise", "target": <name of an existing feature>, "name": ..., "kind": ...,
"question": ...}}
Replace that feature's question with better wording. Use this when a feature measures the
right thing badly: too narrow, too vague, or worded so nearly every note answers the same.
- {{"op": "drop", "target": <name of an existing feature>, "name": "", "kind": "intensity",
"question": ""}}
Remove a feature that is not earning its place.
`kind` is "intensity" for something with a degree, or "presence" for a yes/no fact.
`question` is what gets asked about one tasting note.
An "intensity" question is graded against this fixed five-level rubric, so word it so that the
levels make sense:
{chr(10).join(f" {i}. {level}" for i, level in enumerate(INTENSITY_LEVELS))}
A "presence" question is answered as the probability that it is true of the note.
Good features can be judged from the note's own words, vary from note to note, and carry
information about quality that the other features do not. Reviewers describe structure, fruit,
oak, length, complexity, and drinkability, and they also signal quality through word choice."""
class Split(NamedTuple):
"""The rows, their labels, and which half the loop is allowed to read."""
notes: list[str]
scores: np.ndarray
dev: np.ndarray
test: np.ndarray
# ----------------------------------------------------------------- the data
WINEMAG_CSV = (
"https://huggingface.co/datasets/GroNLP/ik-nlp-22_winemag/resolve/"
"90eb39f35fc64e556fc17f06d4137a4a69ec3297/train.csv"
)
@json_cache
def load_slice(n_dev: int, n_test: int, seed: int) -> dict:
"""Fetch the pinned CSV and take a seeded sample of note + score, one row per note."""
import csv
import io
request = urllib.request.Request(
WINEMAG_CSV, headers={"User-Agent": "typesafe-cookbook/1.0"}
)
with urllib.request.urlopen(request, timeout=300) as response:
text = response.read().decode()
rows, seen = [], set()
for row in csv.DictReader(io.StringIO(text)): # a few notes repeat verbatim
if not row["description"] or not row["points"] or row["description"] in seen:
continue
seen.add(row["description"])
rows.append((row["description"], float(row["points"])))
random.Random(seed).shuffle(rows)
picked = rows[: n_dev + n_test]
return {"notes": [r[0] for r in picked], "points": [r[1] for r in picked]}
def example_rows(split: Split, out_of_fold: np.ndarray | None, n: int) -> list[int]:
"""Select representative dev rows for a proposer round."""
dev = split.dev
if out_of_fold is None:
ranked = dev[np.argsort(split.scores[dev], kind="stable")]
return [int(ranked[round(q * (len(ranked) - 1))]) for q in np.linspace(0, 1, n)]
error = np.abs(split.scores[dev] - out_of_fold)
order = np.argsort(-error, kind="stable")
worst = [int(dev[i]) for i in order[: n // 2]]
best = [int(dev[i]) for i in order[len(order) - (n - n // 2) :]]
return worst + best
def example_block(
rows: list[int],
split: Split,
out_of_fold: np.ndarray | None,
previous: np.ndarray | None = None,
) -> str:
"""Format selected rows for the proposer."""
if out_of_fold is None:
head = "Example notes, with the score each one was given:"
body = [f"- scored {split.scores[r]:.0f}: {split.notes[r]}" for r in rows]
return head + "\n" + "\n".join(body)
head = (
"Dev notes, worst-predicted first. The first half is where your current questions "
"miss by the most and the second half is where they are already right, so what "
"separates the halves is what the questions have not captured."
)
if previous is not None:
head += (
" Each line also carries what the previous round predicted, so you can see which "
"notes your last batch of questions moved."
)
body = []
for r in rows:
line = f"- scored {split.scores[r]:.0f}, predicted {out_of_fold[r]:.1f}"
if previous is not None:
line += f" (last round {previous[r]:.1f})"
body.append(f"{line}: {split.notes[r]}")
return head + "\n" + "\n".join(body)
def load_split(n_dev: int, n_test: int, seed: int = 0) -> Split:
# keyword, because the cache key is the function name plus how each argument was spelled
data = load_slice(n_dev, n_test, seed=seed)
return Split(
notes=data["notes"],
scores=np.array(data["points"]),
dev=np.arange(n_dev),
test=np.arange(n_dev, n_dev + n_test),
)
# ----------------------------------------------------------------- step 1: propose
def proposal_prompt(examples: str, feedback: str, accepted: list[dict]) -> str:
parts = [PROPOSER_TASK, "\n" + examples]
if accepted:
parts.append(
"\nThe features you have now. `add` must not duplicate one of these; `revise` and "
"`drop` refer to one by name:\n"
+ "\n".join(
f"- {f['name']} ({f['kind']}): {f['question']}" for f in accepted
)
)
if feedback:
parts.append("\nHow the model did with those features:\n" + feedback)
return "\n".join(parts)
@json_cache
def propose(model: str, round_index: int, prompt: str) -> dict:
"""One proposal call. Every number in `prompt` is rounded so a replay hits the cache."""
if model.startswith("claude"):
import anthropic
response = anthropic.Anthropic(
api_key=os.environ.get("ANTHROPIC_API_KEY", "cache-only")
).messages.create(
model=model,
max_tokens=16000,
output_config={
"effort": "medium",
"format": {"type": "json_schema", "schema": PROPOSAL_SCHEMA},
},
messages=[{"role": "user", "content": prompt}],
)
body = next(block.text for block in response.content if block.type == "text")
usage = [response.usage.input_tokens or 0, response.usage.output_tokens or 0]
else:
from openai import OpenAI
response = OpenAI(
api_key=os.environ.get("OPENAI_API_KEY", "cache-only")
).chat.completions.create(
model=model,
reasoning_effort="high",
max_completion_tokens=16000,
response_format={"type": "json_object"},
messages=[
{
"role": "user",
"content": prompt
+ "\n\nReply with JSON matching this schema:\n"
+ json.dumps(PROPOSAL_SCHEMA),
}
],
)
body = response.choices[0].message.content
usage = [response.usage.prompt_tokens, response.usage.completion_tokens]
return {"actions": json.loads(body)["actions"][:PROPOSALS], "usage": usage}
def slug(name: str, taken: set[str]) -> str:
"""Names become question ids and column labels, so keep them plain and unique."""
base = (
"".join(c if c.isalnum() else "_" for c in name.lower()).strip("_") or "feature"
)
candidate, n = base, 2
while candidate in taken:
candidate, n = f"{base}_{n}", n + 1
return candidate
def to_candidates(actions: list[dict], accepted: list[dict], round_index: int) -> tuple:
"""Split a round's actions into screenable candidates and a list of names to drop."""
live = {f["name"] for f in accepted}
drops = [a["target"] for a in actions if a["op"] == "drop" and a["target"] in live]
replacing = {
a["target"] for a in actions if a["op"] == "revise" and a["target"] in live
}
# a revision may keep the name it replaces, since that feature is on its way out
taken, candidates = live - replacing, []
for action in actions:
if action["op"] == "drop":
continue
if action["op"] == "revise" and action["target"] not in live:
continue # a revision of something that is not there
name = slug(action["name"], taken)
taken.add(name)
candidates.append(
{
"id": f"{name}@{round_index}", # unique, so earlier rounds keep their columns
"name": name,
"kind": action["kind"],
"question": action["question"],
"replaces": action["target"] if action["op"] == "revise" else "",
}
)
return candidates, drops
# ----------------------------------------------------------------- step 2: answer
def feature_questions(features: list[dict]) -> dict:
questions = {}
for feature in features:
if feature["kind"] == "intensity":
questions[feature["name"]] = Score(
instructions=feature["question"], criteria=INTENSITY_LEVELS
)
else:
questions[feature["name"]] = Noul(
instructions=feature["question"], criteria=PRESENCE_CRITERIA
)
return questions
@json_cache
def answer(model: str, note: str, features_json: str) -> dict:
"""One request per note; every question of the round rides it. Keeps every probability."""
features = json.loads(features_json)
started = perf_counter()
response = client.system_one(
state=note, questions=feature_questions(features), model=model
)
raw = {}
for feature in features:
got = response.answers[feature["name"]]
if feature["kind"] == "intensity":
raw[feature["name"]] = [
got.probabilities.get(i, 0.0) for i in range(len(INTENSITY_LEVELS))
]
else:
raw[feature["name"]] = [got.noul]
return {
"raw": raw,
"seconds": round(perf_counter() - started, 2),
"input_tokens": response.usage.input_tokens or 0,
"output_tokens": response.usage.output_tokens or 0,
}
def featurize(notes: list[str], features: list[dict]) -> dict:
"""Answer one question set for many notes: one request each, eight in flight."""
payload = json.dumps(features, sort_keys=True)
with ThreadPoolExecutor(max_workers=8) as pool:
results = list(
pool.map(lambda note: answer(TYPESAFE_MODEL, note, payload), notes)
)
return {
f["name"]: np.array([r["raw"][f["name"]] for r in results], dtype=float)
for f in features
}
def encode(feature: dict, probabilities: np.ndarray, mode: str) -> list[tuple]:
"""Turn one question's probabilities into named columns."""
name = feature["name"]
if feature["kind"] == "presence":
return [(name, probabilities[:, 0])] # one number is all there is
levels = np.arange(probabilities.shape[1])
mean = probabilities @ levels
if mode == "mean":
return [(name, mean)]
if mode == "mean_spread":
variance = probabilities @ (levels**2) - mean**2
return [(name, mean), (f"{name}_sd", np.sqrt(np.clip(variance, 0, None)))]
return [(f"{name}_p{i}", probabilities[:, i]) for i in levels]
def design(features: list[dict], answers_for: dict, mode: str) -> tuple:
"""Stack every feature's columns into one matrix, plus a label per column."""
columns, labels = [], []
for feature in features:
for label, column in encode(feature, answers_for[feature["id"]], mode):
columns.append(column)
labels.append(label)
return np.column_stack(columns), labels
def plain(features: list[dict]) -> list[dict]:
"""What goes on the wire and into the cache key: no id, no bookkeeping."""
return [
{"name": f["name"], "kind": f["kind"], "question": f["question"]}
for f in features
]
# ----------------------------------------------------------------- step 3: fit
def rmse(y: np.ndarray, p: np.ndarray) -> float:
return float(np.sqrt(np.mean((y - p) ** 2)))
def spearman(a: np.ndarray, b: np.ndarray) -> float:
"""Rank correlation: does the model order the wines the way the critic did?"""
ranks = (
np.argsort(np.argsort(a)).astype(float),
np.argsort(np.argsort(b)).astype(float),
)
return float(np.corrcoef(*ranks)[0, 1])
def folds(y: np.ndarray, k: int, seed: int) -> list[np.ndarray]:
"""Label-stratified k-fold: sort by the label with a seeded tiebreak, then deal off the top."""
rng = np.random.default_rng(seed)
order = np.lexsort((rng.random(len(y)), y))
return [np.sort(order[i::k]) for i in range(k)]
def cross_validate(X: np.ndarray, y: np.ndarray) -> tuple[np.ndarray, float]:
out_of_fold = np.zeros((REPEATS, len(y)))
for repeat in range(REPEATS):
for fold in folds(y, FOLDS, seed=repeat):
train = np.setdiff1d(np.arange(len(y)), fold)
model = CatBoostRegressor(**CATBOOST).fit(X[train], y[train])
out_of_fold[repeat, fold] = model.predict(X[fold])
scores = [rmse(y, out_of_fold[repeat]) for repeat in range(REPEATS)]
return out_of_fold.mean(axis=0), float(np.mean(scores))
def importances(X: np.ndarray, y: np.ndarray) -> np.ndarray:
return CatBoostRegressor(**CATBOOST).fit(X, y).get_feature_importance()
def paired_gain(y: np.ndarray, before: np.ndarray, after: np.ndarray) -> tuple:
"""Bootstrap the paired held-out RMSE change."""
squared = ((y - before) ** 2, (y - after) ** 2)
rng = np.random.default_rng(0)
drawn = []
for _ in range(2000):
rows = rng.integers(0, len(y), len(y))
drawn.append(
np.sqrt(squared[1][rows].mean()) - np.sqrt(squared[0][rows].mean())
)
drawn = np.array(drawn)
return (
rmse(y, after) - rmse(y, before),
float(np.percentile(drawn, 2.5)),
float(np.percentile(drawn, 97.5)),
)
def fit_predict(X: np.ndarray, split: Split) -> np.ndarray:
model = CatBoostRegressor(**CATBOOST).fit(X[split.dev], split.scores[split.dev])
return model.predict(X[split.test])
def fit_predict_text(split: Split) -> np.ndarray:
"""The reference arm: the same model, handed the note instead of the columns."""
from catboost import Pool
raw = np.array([[note] for note in split.notes], dtype=object)
model = CatBoostRegressor(**CATBOOST).fit(
Pool(raw[split.dev], split.scores[split.dev], text_features=[0])
)
return model.predict(Pool(raw[split.test], text_features=[0]))
def evaluate(
features: list[dict], answers_for: dict, split: Split, mode: str
) -> tuple[np.ndarray, float]:
"""Cross-validated error on the dev rows for one candidate question set."""
X, _ = design(features, answers_for, mode)
return cross_validate(X[split.dev], split.scores[split.dev])
def swap_in(accepted: list[dict], feature: dict) -> list[dict] | None:
"""The accepted set with `feature` in place of the one it revises, or None if it is gone."""
at = next(
(i for i, f in enumerate(accepted) if f["name"] == feature["replaces"]), None
)
if at is None:
return None
trial = list(accepted)
trial[at] = {k: feature[k] for k in ("id", "name", "kind", "question")}
return trial
def try_change(
trial: list[dict],
accepted: list[dict],
cv: float,
answers_for: dict,
split: Split,
mode: str,
tolerance: float,
) -> tuple[list[dict], float, str, bool]:
"""Refit with the change and keep it only if the dev error improves. No API calls."""
_, cv_trial = evaluate(trial, answers_for, split, mode)
if cv_trial <= cv + tolerance:
return trial, cv_trial, f"CV {cv:.3f} -> {cv_trial:.3f}", True
return accepted, cv, f"would cost {cv_trial - cv:+.3f}", False
def owner_of(label: str, features: list[dict]) -> dict:
"""Which feature a column label belongs to - encodings suffix the name."""
exact = next((f for f in features if f["name"] == label), None)
if exact:
return exact
return next(f for f in features if label.startswith(f["name"] + "_"))
def importance_per_feature(
features: list[dict], labels: list[str], column_importances: np.ndarray
) -> dict:
"""Sum each question's CatBoost column importances.
Intensity questions can produce multiple model columns. Combining their normalized
importances gives one percentage share per question.
"""
total = {f["name"]: 0.0 for f in features}
for label, column_importance in zip(labels, column_importances):
total[owner_of(label, features)["name"]] += float(column_importance)
return total
def feedback_for(
history: list[float],
accepted: list[dict],
answers_for: dict,
split: Split,
mode: str,
out_of_fold: np.ndarray,
previous: np.ndarray | None,
) -> str:
"""The scoreboard the next proposal call reads. The notes themselves arrive separately,
through `example_block`. Numbers are rounded before they enter the prompt."""
X, labels = design(accepted, answers_for, mode)
dev, scores = split.dev, split.scores
by_name = importance_per_feature(accepted, labels, importances(X[dev], scores[dev]))
lines = ["Cross-validated RMSE in points so far, lower is better:"]
lines += [f" round {i + 1}: {v:.2f}" for i, v in enumerate(history)]
if previous is not None:
now, before = np.abs(scores[dev] - out_of_fold), np.abs(scores[dev] - previous)
better, worse = int((now < before - 0.1).sum()), int((now > before + 0.1).sum())
lines.append(
f"\nAgainst the previous round, {better} of the {len(dev)} dev notes are now "
f"predicted better by more than 0.1 points and {worse} are predicted worse."
)
lines.append(
"\nYour features, with importance as a percentage of the total and the spread of the "
"column across the dev rows. Low importance or low spread means the question is not "
"doing much; revise or drop it."
)
for feature in sorted(accepted, key=lambda f: -by_name.get(f["name"], 0.0)):
column = encode(feature, answers_for[feature["id"]], mode)[0][1]
lines.append(
f" {feature['name']} ({feature['kind']}): "
f"{by_name.get(feature['name'], 0.0):.1f}% importance, "
f"spread {column[dev].std():.2f}"
)
return "\n".join(lines)
# ----------------------------------------------------------------- the loop itself
class Discovery(NamedTuple):
"""Artifacts returned by the discovery loop."""
accepted: list[dict] # the question set it ended with
answers_for: dict # feature id -> (rows x levels) probabilities
snapshots: list[list[dict]] # the set as it stood at the end of each round
history: list[float] # dev CV error after each round
batches: list[tuple] # what each round sent, for the request table
journal: list[tuple] # every action and what became of it
def run_loop(
split: Split,
proposer: str,
rounds: int,
examples: int,
mode: str,
min_spread: float,
tolerance: float,
) -> Discovery:
"""Run the propose, answer, fit, and feedback loop."""
shown = example_rows(split, None, examples) # round 1 has nothing predicted yet
out_of_fold = previous = None
got_from = Discovery([], {}, [], [], [], [])
accepted, answers_for = got_from.accepted, got_from.answers_for
snapshots, history = got_from.snapshots, got_from.history
batches, journal = got_from.batches, got_from.journal
feedback = ""
for round_index in range(1, rounds + 1):
block = example_block(shown, split, out_of_fold, previous)
actions = propose(
proposer, round_index, proposal_prompt(block, feedback, accepted)
)["actions"]
keep, drops = to_candidates(actions, accepted, round_index)
if keep: # one request per row, carrying every question this round proposed
batches.append((round_index, plain(keep)))
answers = featurize(split.notes, plain(keep))
for feature in keep:
answers_for[feature["id"]] = answers[feature["name"]]
for (
feature
) in keep: # an add goes in; importance says later whether it earned it
if feature["replaces"]:
continue
column = encode(feature, answers_for[feature["id"]], mode)[0][1]
flat = float(column[split.dev].std()) < min_spread
journal.append(
(round_index, "flat" if flat else "add", feature["name"], "")
)
if not flat:
accepted.append(
{k: feature[k] for k in ("id", "name", "kind", "question")}
)
_, cv = evaluate(accepted, answers_for, split, mode)
trial_args = (answers_for, split, mode, tolerance)
for feature in [f for f in keep if f["replaces"]]: # every revision is tried
trial = swap_in(accepted, feature)
if trial is None: # it revises something an earlier round already dropped
journal.append(
(round_index, "stale", feature["name"], "target is gone")
)
continue
accepted[:], cv, note, took = try_change(trial, accepted, cv, *trial_args)
what = "revise" if took else "reject"
journal.append(
(
round_index,
what,
feature["name"],
f"was {feature['replaces']}, {note}",
)
)
for name in drops: # and so is every drop
trial = [f for f in accepted if f["name"] != name]
if not trial:
continue
accepted[:], cv, note, took = try_change(trial, accepted, cv, *trial_args)
journal.append((round_index, "drop" if took else "keep", name, note))
previous, (out_of_fold, cv) = (
out_of_fold,
evaluate(accepted, answers_for, split, mode),
)
history.append(cv)
snapshots.append(list(accepted))
feedback = feedback_for(
history, accepted, answers_for, split, mode, out_of_fold, previous
)
# next round reads the rows these questions get most wrong, and as many they get right
shown = example_rows(split, out_of_fold, examples)
report(round_index, keep, drops, journal, accepted, cv)
return got_from
def report(
round_index: int,
keep: list[dict],
drops: list[str],
journal: list[tuple],
accepted: list[dict],
cv: float,
) -> None:
"""One block per round: the counts, the names it added, then everything with a number."""
revised = sum(1 for f in keep if f["replaces"])
print(
f"round {round_index}: {len(keep) - revised} add, {revised} revise, "
f"{len(drops)} drop"
)
this_round = [j for j in journal if j[0] == round_index]
added = [name for _, what, name, _ in this_round if what == "add"]
if added:
print(
textwrap.fill(
", ".join(added),
88,
initial_indent=" added ",
subsequent_indent=" " * 10,
)
)
for _, what, name, note in this_round: # everything carrying a number of its own
if what != "add":
print(f" {what:<7}{name:<34}{note}")
print(f" -> {len(accepted)} features, dev CV RMSE {cv:.3f}\n")
# ----------------------------------------------------------------- asking for the score
@json_cache
def ask_score(model: str, note: str) -> dict:
"""One `Score` over ten quality bands, read as a level and rescaled to 80-100."""
response = client.system_one(
state=note,
questions={
"quality": Score(
instructions=(
"Judging only by what this tasting note says, how good is the wine?"
),
criteria=SCORE_LEVELS,
)
},
model=model,
)
got = response.answers["quality"]
top = len(SCORE_LEVELS) - 1
expected = sum(k * v for k, v in got.probabilities.items())
return {
# level 0 is the bottom of the critic's scale, level 9 the top
"expected": 80.0 + 20.0 * expected / top,
"picked": 80.0 + 20.0 * got.score / top,
"input_tokens": response.usage.input_tokens or 0,
"output_tokens": response.usage.output_tokens or 0,
}
# ----------------------------------------------------------------- charts
SURFACE, INK, INK2, MUTED = "#fcfcfb", "#0b0b0b", "#52514e", "#898781"
GRID, AXIS, BLUE, ORANGE = "#e1e0d9", "#c3c2b7", "#2a78d6", "#eb6834"
def style(ax) -> None:
ax.set_facecolor(SURFACE)
for side in ("top", "right"):
ax.spines[side].set_visible(False)
for side in ("left", "bottom"):
ax.spines[side].set_color(AXIS)
ax.tick_params(colors=MUTED, labelcolor=INK2, labelsize=9)
ax.set_axisbelow(True)
def polarity(feature: dict, answers_for: dict, split: Split) -> float:
"""Rank correlation between a question's answer and the critic score, on the dev rows.
Positive means a higher answer goes with a better review, negative the opposite. It is
what orders the rows of the feature map, so the map reads as a gradient that flips.
"""
column = encode(feature, answers_for[feature["id"]], "mean")[0][1]
return spearman(column[split.dev], split.scores[split.dev])
def reviews_heatmap(
plt,
questions: list[dict],
answers_for: dict,
split: Split,
rows: tuple,
):
"""Compare held-out reviews across the discovered questions, best-signal first.
Rows arrive sorted from the questions that rise with the score to the ones that fall with
it, so a row above the divider shades left to right and a row below it shades right to
left.
"""
def value_of(feature: dict, row: int) -> float:
return float(encode(feature, answers_for[feature["id"]], "mean")[0][1][row])
signs = [polarity(question, answers_for, split) for question in questions]
flip = next((i for i, s in enumerate(signs) if s < 0), len(questions))
raw = np.array(
[[value_of(question, row) for row in rows] for question in questions]
)
normalized = np.array(
[
values / (4 if question["kind"] == "intensity" else 1)
for question, values in zip(questions, raw)
]
)
cmap = matplotlib.colors.LinearSegmentedColormap.from_list(
"typesafe_heat", [SURFACE, "#f7c7ad", ORANGE]
)
fig, ax = plt.subplots(
figsize=(9.5, 1.8 + 0.58 * len(questions)), facecolor=SURFACE
)
image = ax.imshow(normalized, aspect="auto", cmap=cmap, vmin=0, vmax=1)
row_labels = []
for question, sign in zip(questions, signs):
kind = "score" if question["kind"] == "intensity" else "noul"
prefix = f"{sign:+.2f} ({kind}) "
lines = textwrap.wrap(
" ".join(question["question"].split()),
width=52,
max_lines=2,
placeholder="...",
break_long_words=False,
break_on_hyphens=False,
)
row_labels.append(prefix + (f"\n{' ' * len(prefix)}").join(lines))
column_labels = [
f"#{i}\n{split.scores[row]:.0f} points\n{' '.join(split.notes[row].split())[:15]}..."
for i, row in enumerate(rows, 1)
]
ax.set_yticks(np.arange(len(questions)), row_labels)
ax.set_xticks(np.arange(len(rows)), column_labels)
ax.tick_params(
axis="x", top=True, labeltop=True, bottom=False, labelbottom=False, pad=8
)
ax.tick_params(axis="y", labelsize=8.5)
for side in ax.spines.values():
side.set_visible(False)
ax.set_xticks(np.arange(-0.5, len(rows), 1), minor=True)
ax.set_yticks(np.arange(-0.5, len(questions), 1), minor=True)
ax.grid(which="minor", color=SURFACE, linewidth=2)
ax.tick_params(which="minor", bottom=False, left=False)
for i, question in enumerate(questions):
for j, value in enumerate(raw[i]):
label = (
f"{value:.1f}" if question["kind"] == "intensity" else f"{value:.2f}"
)
color = SURFACE if normalized[i, j] > 0.58 else INK2
ax.text(j, i, label, ha="center", va="center", color=color, fontsize=8)
# the line where the questions stop rising with the score and start falling with it
if 0 < flip < len(questions):
ax.axhline(flip - 0.5, color=INK, linewidth=1.2)
ax.annotate(
"a higher answer means a worse review, below this line",
(len(rows) - 0.5, flip - 0.5),
xytext=(-4, 5),
textcoords="offset points",
va="bottom",
ha="right",
color=INK2,
fontsize=8.5,
)
colorbar = fig.colorbar(image, ax=ax, fraction=0.025, pad=0.025)
colorbar.set_ticks([0, 0.5, 1])
colorbar.set_label("normalized answer", color=INK2, fontsize=8.5)
colorbar.ax.tick_params(labelsize=8, colors=INK2)
fig.suptitle(
"Every question, on five held-out reviews from worst to best",
x=0.01,
y=0.995,
ha="left",
color=INK,
fontsize=11,
)
fig.text(
0.01,
0.972,
"sorted by how the answer moves with the score, so each row above the line shades "
"left to right and each row below it shades the other way",
color=MUTED,
fontsize=9,
)
fig.text(
0.01,
0.005,
"Row labels lead with the rank correlation between that question's answer and the "
"critic score. Cell text is each question's native scale: score 0-4, noul 0-1.",
color=MUTED,
fontsize=8.5,
)
return fig
def rounds_chart(
plt, curve: list[tuple], history: list[float], n_test: int, gain: tuple
):
"""Dev error and held-out error per round. The trend is the point, not the gap."""
rounds = list(range(1, len(curve) + 1))
values = [v for _, v in curve]
fig, ax = plt.subplots(figsize=(7, 3.9), facecolor=SURFACE)
style(ax)
ax.grid(axis="y", color=GRID, linewidth=0.8)
# each dev fold trains on four fifths of the rows, so the dev line sits the higher of the two
ax.fill_between(rounds, history, values, color=GRID, alpha=0.75, linewidth=0)
ax.plot(
rounds,
history,
marker="o",
color=BLUE,
linewidth=2,
linestyle="--",
label="dev, cross-validated - what the loop optimises",
)
ax.plot(
rounds,
values,
marker="o",
color=ORANGE,
linewidth=2,
label="held out - what that actually buys",
)
# label each point on the outside of the pair, so neither line crowds its own numbers
for x, dev_value, test_value in zip(rounds, history, values):
for value, other in ((dev_value, test_value), (test_value, dev_value)):
ax.annotate(
f"{value:.2f}",
(x, value),
textcoords="offset points",
xytext=(0, 8 if value >= other else -16),
ha="center",
color=INK2,
fontsize=8.5,
)
ax.set_xticks(
rounds, [f"round {x}\n{n} features" for x, (n, _) in zip(rounds, curve)]
)
ax.set_ylabel("RMSE in points (lower is better)", color=INK2, fontsize=9)
# tight around the two lines: the whole finding lives inside 0.15 of a point
low, high = min(values + history), max(values + history)
ax.set_ylim(low - 0.10, high + 0.05)
difference, low_ci, high_ci = gain
ax.set_title(
f"{len(rounds)} rounds of the loop, scored on {n_test} held-out reviews",
loc="left",
color=INK,
fontsize=11,
pad=20,
)
# the number the chart is really about: is the held-out move bigger than the noise?
ax.text(
0,
1.015,
f"round 1 to round {len(rounds)}, held out: {difference:+.3f} points, "
f"95% CI [{low_ci:+.3f}, {high_ci:+.3f}]",
transform=ax.transAxes,
color=MUTED,
fontsize=9,
)
ax.legend(frameon=False, labelcolor=INK2, fontsize=9, loc="lower left")
return fig설정
pip install anthropic openai catboost numpy matplotlib ipython 'cooksafe>=0.2.0,<0.3.0'그런 다음 TYPESAFE_API_KEY와 ANTHROPIC_API_KEY를 설정합니다. 모든 API 호출은 cookbook과 함께 제공되는 json_cache.json에 캐시되므로, 다시 렌더링하면 아무것도 호출하지 않고 이 수치들을 재생합니다. 파일을 삭제하면 실시간으로 다시 실행합니다. 수치는 2026-08-03에 TypeSafe jev-1.12와 claude-sonnet-5에서 나왔습니다. propose()에는 gpt-5.6-luna용 두 번째 분기가 있는데, 실행하지 않았습니다.
첫 번째 코드 셀은 구현 전체입니다. API 호출, 인코딩, 지표, 차트 스타일입니다. 이 파일이 단독으로 실행되도록 넣어 둔 것이며, 문서 사이트에서는 접어 둡니다. 처음 읽을 때는 건너뛰십시오. 레시피는 그 아래에서 시작합니다.
N_DEV, N_TEST = 1200, 800 # the loop reads dev labels only; test is scored once
ROUNDS = 5 # a round answers questions for all 2,000 rows: 2,000 requests
PROPOSER = "claude-sonnet-5" # or "gpt-5.6-luna"; the cache holds the Anthropic run
EXAMPLES = 60 # dev notes the proposer reads per round, half of them its worst misses
MIN_SPREAD = 0.05 # a column this flat cannot separate anything, so it is not kept
CHANGE_TOLERANCE = 0.0 # a revision or drop has to improve dev error, not just not hurt
ENCODING = "mean_spread" # a score answer becomes two columns: its mean and spread
split = load_split(N_DEV, N_TEST, seed=0)
NOTES, SCORES, DEV, TEST = split.notes, split.scores, split.dev, split.test
print(
f"{len(DEV)} dev rows, {len(TEST)} held out; scores run "
f"{SCORES.min():.0f}-{SCORES.max():.0f}, mean {SCORES.mean():.2f}, sd {SCORES.std():.2f}"
)
print(f"\none of the notes:\n{NOTES[0]}")1200 dev rows, 800 held out; scores run 80-98, mean 88.73, sd 3.17
one of the notes:
A Champagne that is very much wine. The structure and the richness are just right for a food wine, showing ripe acidity, flavors of plums and apricots, and balancing these primary fruits with a dense, complex structure that takes in yeast, maturity and a tight apple skin finish.루프는 2,000개 행 중 같은 1,200개(dev 행)를 반복해서 읽고, 그 1,200개 점수를 예측하는 데 도움이 되는 질문을 남깁니다. 같은 행에서 채점하면 대부분 루프가 그 행들에 얼마나 잘 맞춰졌는지를 측정하게 되므로, 나머지 800개는 홀드아웃으로 두고 마지막에 한 번 채점합니다.
두 가지 질문 유형
제안된 질문은 두 종류 중 하나이며, 그 종류가 어떤 숫자가 돌아오는지를 결정합니다.
- **
intensity**는 정도가 있는 모든 것에 대해Score가 됩니다. 그 다섯 레벨은 아래에 출력되며, 열은 평균 레벨이므로 “moderate”와 “strongly” 사이에 있는 노트는 둘 사이 값으로 나옵니다. - **
presence**는 결함이 언급되었는지 같은 예/아니오 사실에 대해Noul이 됩니다. 열은 그 확률 하나입니다.
방법
questions <- {}
repeat for each round:
notes <- round 1 ? 60 dev notes across the score range
: the 30 worst-predicted dev notes + the 30 best,
each with its score, this prediction and the last
actions <- LLM(brief, questions, notes, importance and error so far)
answers[q] <- TypeSafe(note, all new questions of this round) for every row
for each added q: keep it unless its column is flat
for each revised q: refit; keep the change only if dev error drops
for each dropped q: refit; drop it only if dev error drops
out_of_fold <- k-fold CatBoost on the columns # judges, and picks next round's notes어떤 질문도 답하기 전에 걸러지지 않습니다. 한 라운드의 모든 질문은 같은 요청에 실려 나가므로, 질문이 하나 더 있어도 요청이 늘지 않습니다. 열 행 중 한 행에만 해당하는 질문은 제안자가 읽는 60개 노트에서는 쓸모없어 보이지만, 집합에서 가장 유용한 열일 수 있습니다.
k-fold는 dev 행을 k개 부분으로 나누고, 각 부분을 나머지 부분으로 학습한 모델로 예측하는 것입니다. 그 예측은 세 가지 일을 합니다. 모든 수정과 삭제를 심사하고, 다음 라운드가 읽을 노트를 고르며, 제안자에게 자기 질문 중 무엇이 도움이 되었는지, 이전 라운드 이후 얼마나 움직였는지로 알려줍니다.
print("every intensity question is graded on these five levels:\n")
for i, level in enumerate(INTENSITY_LEVELS):
print(f" {i}. {level}")
print("\nevery presence question is judged true or false against these:\n")
print(f" true: {PRESENCE_CRITERIA['true']}")
print(f" false: {PRESENCE_CRITERIA['false']}")
print("\nthe brief the proposer works from:\n")
print("\n".join(PROPOSER_TASK.splitlines()[:6]) + "\n ...")every intensity question is graded on these five levels:
0. Not present in this note at all
1. Barely present - mentioned once, in passing
2. Present at a moderate level
3. Present strongly - the note dwells on it
4. Dominant - the note is largely about this
every presence question is judged true or false against these:
true: The note states this or clearly implies it
false: The note gives no indication of this
the brief the proposer works from:
You are designing numeric features for a gradient-boosting model that
predicts the score a wine critic gave (an integer from 80 to 100) from the tasting note alone.
The model sees nothing but the features you design.
Return up to 18 actions. Each action is one of:
...autoresearch 루프
run_loop는 다섯 라운드를 모두 실행하고 라운드마다 블록 하나를 출력합니다. 추가된 질문은 곧바로 들어갑니다. 그 답은 이미 가져왔고, 중요도가 나중에 물어볼 가치가 있었는지를 보여줍니다. 수정이나 삭제는 모델이 이미 쓰고 있는 열을 없애므로, 각각을 먼저 시도합니다. 변경을 넣고 다시 적합시켜, dev 오류가 내려갈 때만 유지합니다. 다시 적합시키는 데는 API 호출이 들지 않으므로, 변경을 시도하고 거부하는 것은 무료입니다.
run = run_loop(
split, PROPOSER, ROUNDS, EXAMPLES, ENCODING, MIN_SPREAD, CHANGE_TOLERANCE
)
accepted, answers_for = run.accepted, run.answers_for
snapshots, history = run.snapshots, run.historyround 1: 18 add, 0 revise, 0 drop
added complexity, fruit_intensity, tannin_structure, acidity_intensity,
oak_intensity, finish_length, balance_harmony, aging_potential,
positive_superlative_language, negative_critical_language,
drinkability_easiness, body_richness, sweetness_level, texture_descriptors,
earthy_savory_notes, flaw_or_defect_mentioned,
single_vineyard_or_prestige_signal, varietal_blend_detail
-> 18 features, dev CV RMSE 1.903
round 2: 5 add, 3 revise, 3 drop
added power_concentration_language, flavor_distinctiveness, generic_fruit_language,
candied_artificial_flavor, rustic_authentic_character
reject oak_dominance was oak_intensity, would cost +0.005
revise negative_critical_language was negative_critical_language, CV 1.897 -> 1.894
revise single_vineyard_or_prestige_signalwas single_vineyard_or_prestige_signal, CV 1.894 -> 1.881
keep finish_length would cost +0.009
keep texture_descriptors would cost +0.001
keep varietal_blend_detail would cost +0.023
-> 23 features, dev CV RMSE 1.881
round 3: 7 add, 2 revise, 1 drop
added elegance_finesse_language, minerality_precision_language,
hedged_qualified_praise, underripe_green_character,
reviewer_overall_verdict_strength, unusual_or_funky_descriptor_valence,
botrytis_or_special_winemaking_signal
revise negative_critical_language was negative_critical_language, CV 1.868 -> 1.864
revise finish_quality was finish_length, CV 1.864 -> 1.861
keep candied_artificial_flavor would cost +0.014
-> 30 features, dev CV RMSE 1.861
round 4: 5 add, 2 revise, 3 drop
added excess_or_imbalance_signal, descriptive_detail_density,
critic_enthusiasm_confidence, savory_food_wine_seriousness,
note_overall_tone_positivity
revise rustic_authentic_character was rustic_authentic_character, CV 1.843 -> 1.838
reject hedged_qualified_praise was hedged_qualified_praise, would cost +0.014
keep botrytis_or_special_winemaking_signalwould cost +0.011
keep candied_artificial_flavor would cost +0.009
keep unusual_or_funky_descriptor_valencewould cost +0.010
-> 35 features, dev CV RMSE 1.838
round 5: 4 add, 2 revise, 8 drop
added structural_seriousness, youthful_tension_signal, surface_prettiness_vs_depth,
price_value_signal
reject unconventional_character_as_virtuewas rustic_authentic_character, would cost +0.010
revise flavor_distinctiveness was flavor_distinctiveness, CV 1.849 -> 1.843
keep candied_artificial_flavor would cost +0.002
keep botrytis_or_special_winemaking_signalwould cost +0.003
keep hedged_qualified_praise would cost +0.006
keep excess_or_imbalance_signal would cost +0.005
drop underripe_green_character CV 1.843 -> 1.840
keep unusual_or_funky_descriptor_valencewould cost +0.002
keep texture_descriptors would cost +0.001
keep generic_fruit_language would cost +0.000
-> 38 features, dev CV RMSE 1.840자신의 데이터에 겨누기
PROPOSER_TASK는 와인을 언급하는 유일한 문자열이며, featurize()는 임의의 문자열 목록을 받습니다. 그 브리프를 고치면 제안 프롬프트가 바뀌고, 프롬프트가 캐시 키의 일부이므로, 다음 실행은 라운드마다 API를 다시 호출합니다.
요청 수는 질문이 아니라 행에 따라 늘어납니다. 라운드마다 행당 요청 하나이므로, 100,000행이면 라운드당 100,000 요청입니다. 수정은 새 질문으로 세므로, 모든 행을 한 번 더 훑는 비용이 듭니다. 워커 풀은 천천히 올리십시오. 공유 키에서는 여덟 개만으로도 레이트 리밋에 걸립니다.
질문이 보는 것
홀드아웃 리뷰 다섯 건, 점수 범위를 사등분한 각 지점에서 하나씩, 38개 질문 중 열다섯 개와 대조합니다. 중요도 기준 상위 여덟 개 score 질문에 상위 일곱 개 noul을 더한 것입니다.
그런 다음 그 열다섯 행을 답이 평론가 점수와 함께 어느 방향으로 움직이는지로 정렬합니다. 답이 점수와 함께 오르는 질문이 먼저 오고, 점수와 함께 내리는 질문이 구분선 아래에 옵니다. 따라서 왼쪽에서 오른쪽으로, 최악의 리뷰에서 최고의 리뷰로 가면서, 구분선 위의 답은 올라가고 아래의 답은 떨어져야 합니다.
X, labels = design(accepted, answers_for, ENCODING)
column_importances = importances(X[DEV], SCORES[DEV])
# an encoding gives a feature more than one column, so add a feature's columns back up
feature_importances = importance_per_feature(accepted, labels, column_importances)
ranked = sorted(accepted, key=lambda f: -feature_importances[f["name"]])
score_questions = [f for f in ranked if f["kind"] == "intensity"][:8]
noul_questions = [f for f in ranked if f["kind"] == "presence"][:7]
# ordered by which way the answer moves with the score, so the map flips halfway down
heatmap_questions = sorted(
score_questions + noul_questions,
key=lambda f: -polarity(f, answers_for, split),
)
ordered_test = TEST[np.argsort(SCORES[TEST], kind="stable")]
positions = np.linspace(0, len(ordered_test) - 1, 5).round().astype(int)
review_rows = tuple(ordered_test[positions])
print("the five held-out heatmap columns:\n")
for i, row in enumerate(review_rows, 1):
excerpt = " ".join(NOTES[row].split())
print(f" {i}. {SCORES[row]:.0f} points: {excerpt[:100]}...")
fig = reviews_heatmap(plt, heatmap_questions, answers_for, split, review_rows)
display(fig)
plt.close(fig)the five held-out heatmap columns:
1. 80 points: Raw cherry and plum aromas are resiny and suggest wet cement. This is shearing and so jacked up with...
2. 86 points: A slight spritz brightens the mouthfeel of this lemony wine. Aromas are a bit musky, but flavors of ...
3. 89 points: This is a European-style Syrah, cofermented with 2% Viognier. It's soft and round, medium in body, a...
4. 91 points: From the producer's dry-farmed estate vineyard, and supported by small amounts of Merlot and Caberne...
5. 97 points: A thoroughly elegant, serious and yet immensely enjoyable wine that stays lively many days after ope...
페이지 위쪽의 표를 계산한 것입니다. 다섯 갈래 모두 같은 800개 홀드아웃 행에서 한 번 채점하고, 처음 세 갈래는 특징 발견을 건너뜁니다. 하나는 dev 점수의 평균을 예측하며 노트에서 아무것도 읽지 않습니다. 하나는 text_features 처리를 통해 같은 CatBoost에 노트를 넘기며, 이것이 노트를 단어 수로 바꿉니다. 하나는 TypeSafe에 점수 자체를 요청합니다.
그 세 번째는 행마다 “faulty or unpleasant”부터 “profound”까지 열 개 품질 구간에 걸친 Score 하나입니다. 열 개인 이유는 Score 질문이 받는 레벨은 최대 열 개이고, 열한 개는 서버 오류로 돌아오기 때문입니다. 레벨 0은 80점, 레벨 9는 100점에 대응합니다. 구간을 그렇게 척도에 펼치는 것만으로는 충분하지 않은데, 이 질문에는 이 매체의 점수가 실제로 척도 어디에 있는지가 없기 때문입니다. 그래서 모든 답을 dev 점수에서 측정한 단일 오프셋만큼 옮깁니다. 그 오프셋은 행 레이블에 출력되며, 이 지름길이 점수에서 배우는 유일한 것입니다.
Spearman은 순위 상관으로, 1.0이면 홀드아웃 와인을 평론가의 순서와 정확히 같게 놓은 것입니다. 단어 수 행은 튜닝된 텍스트 회귀 파이프라인이 아니라 CatBoost 자체의 텍스트 처리입니다. 이 모든 것은 데이터셋 하나와 루프 실행 한 번의 결과입니다.
predicted = fit_predict(X, split)
text_predicted = fit_predict_text(split)
# ask TypeSafe for the score itself, one request per row
with ThreadPoolExecutor(max_workers=8) as pool:
direct = list(pool.map(lambda note: ask_score(TYPESAFE_MODEL, note), NOTES))
asked = np.array([d["expected"] for d in direct])
shift = float(SCORES[DEV].mean() - asked[DEV].mean()) # one number, from the dev labels
# what one proposal call gets you, before any feedback: the set round 1 ended with
first_round, _ = design(snapshots[0], answers_for, ENCODING)
print(f"{'arm':<46}{'RMSE':>7}{'spearman':>10}")
for label, p in (
("predict the mean of the dev rows", np.full(len(TEST), SCORES[DEV].mean())),
("the note as word counts, same CatBoost", text_predicted),
(f"ask for the score itself, shifted {shift:+.2f}", asked[TEST] + shift),
(
f"{len(snapshots[0])} questions from round 1, no loop",
fit_predict(first_round, split),
),
(f"{len(accepted)} questions after all {ROUNDS} rounds", predicted),
):
print(f"{label:<46}{rmse(SCORES[TEST], p):>7.3f}{spearman(SCORES[TEST], p):>10.3f}")arm RMSE spearman
predict the mean of the dev rows 3.088 -0.014
the note as word counts, same CatBoost 2.466 0.605
ask for the score itself, shifted -1.71 2.145 0.761
18 questions from round 1, no loop 1.869 0.778
38 questions after all 5 rounds 1.772 0.799autoresearch 라운드가 도움이 되었는가?
두 선 모두 첫 제안에서 시작해 라운드 끝마다 질문 집합의 오류를 그립니다. 점선은 교차 검증된 dev 오류로, 모든 수락과 거부 결정이 이 숫자로 내려집니다. 실선은 같은 질문 집합을 루프가 한 번도 읽지 않는 홀드아웃 행에서 채점합니다. 각 점은 라운드가 닫힐 때의 집합이므로, 질문을 수정하거나 삭제하기만 한 라운드도 두 선을 움직입니다. 특징 지도는 질문이 무엇을 측정하는지 말하고, 오류는 첫 제안 이후의 라운드가 예측을 더 낫게 했는지를 알려줍니다.
축 범위는 좁습니다. 축 위의 모든 일이 0.2점 안에서 벌어지고, 위 표의 모든 지름길은 축 위쪽에서 멀찍이 벗어나 있습니다. dev 선은 내내 홀드아웃 선 위에서 달리는데, 이는 학습 크기 효과입니다. 각 dev 폴드는 dev 행의 5분의 4로 학습하는 반면, 홀드아웃 숫자는 1,200개 전부를 받은 모델에서 나옵니다. 두 선은 함께 움직이므로, 루프가 조타하는 dev 숫자는 결코 보지 못하는 홀드아웃 숫자를 따라갑니다. 제목 아래의 구간은 홀드아웃 행을 재표집한 데서 나오므로, 라운드 1에서 라운드 5로의 이동이 800행의 잡음보다 큰지를 말해줍니다.
curve, per_round = [], []
for features in snapshots:
X_round, _ = design(features, answers_for, ENCODING)
per_round.append(fit_predict(X_round, split))
curve.append((len(features), rmse(SCORES[TEST], per_round[-1])))
# the same held-out rows resampled 2,000 times, both arms scored on each resample
gain = paired_gain(SCORES[TEST], per_round[0], per_round[-1])
print(
f"round 1 -> round {ROUNDS} on the held-out rows: {gain[0]:+.3f} points, "
f"95% CI [{gain[1]:+.3f}, {gain[2]:+.3f}]"
)
fig = rounds_chart(plt, curve, history, len(TEST), gain)
display(fig)
plt.close(fig)round 1 -> round 5 on the held-out rows: -0.097 points, 95% CI [-0.147, -0.050]
홀드아웃 선이 dev 선보다 더 많이 내려갑니다. 라운드 1은 참고할 피드백 없이 질문을 썼고, 그 뒤 네 라운드는 홀드아웃 행에서 0.10점의 가치가 있으며, 95% CI [-0.147, -0.050]입니다.
라운드 5는 추가 네 개, 표현 수정 두 개, 삭제 여덟 개를 제안했고, 처음으로 개선되지 않은 dev 숫자를 냈습니다. 245자 노트에 대해 물을 수 있는 것은 한계가 있고, 라운드 5에 이르러 제안은 질문을 추가하는 쪽에서 삭제하는 쪽으로 기울었습니다.
kinds = {f["name"]: f["kind"] for f in accepted}
print("feature importance share: % of total CatBoost importance across all questions")
print(f"{'feature':<38}{'asked as':<10}{'importance share':>16}")
for name, importance_share in sorted(feature_importances.items(), key=lambda p: -p[1])[
:12
]:
kind = "score" if kinds[name] == "intensity" else "noul"
print(
f"{name[:36]:<38}{kind:<10}{importance_share:>8.1f}% "
f"{'#' * round(importance_share)}"
)
counts = f"{sum(1 for k in kinds.values() if k == 'intensity')} score"
counts += f", {sum(1 for k in kinds.values() if k == 'presence')} noul"
print(f"\nthe {len(accepted)} questions the loop kept: {counts}")
top = max(feature_importances, key=feature_importances.get)
print(
f'the question behind the top row:\n {top}: "{owner_of(top, accepted)["question"]}"'
)feature importance share: % of total CatBoost importance across all questions
feature asked as importance share
note_overall_tone_positivity score 17.4% #################
savory_food_wine_seriousness score 8.7% #########
positive_superlative_language score 8.4% ########
single_vineyard_or_prestige_signal noul 7.2% #######
descriptive_detail_density score 5.7% ######
elegance_finesse_language score 5.0% #####
complexity score 5.0% #####
aging_potential score 5.0% #####
balance_harmony score 2.9% ###
drinkability_easiness score 2.9% ###
critic_enthusiasm_confidence score 2.7% ###
flavor_distinctiveness score 2.6% ###
the 38 questions the loop kept: 29 score, 9 noul
the question behind the top row:
note_overall_tone_positivity: "Setting aside specific descriptors, how positive is the overall emotional tone and word choice of the note taken as a whole (warm, admiring language throughout vs. flat, neutral, or lukewarm phrasing)?"importance share는 CatBoost 특징 중요도로, 38개 질문이 모두 합쳐 100%가 되도록 정규화한 것입니다. 행이나 질문, 예측 정확도의 비율이 아닙니다. score 질문은 평균과 퍼짐이라는 두 열을 가지므로, 백분율을 출력하기 전에 두 열의 중요도를 다시 합칩니다. note_overall_tone_positivity가 전체의 17.4%를 차지합니다. 네 번째 행은 noul입니다. 노트가 단일 밭이나 다른 명성 신호를 언급하는지는 예/아니오 사실이므로, 그렇게 질문했습니다.
다음 단계
이번 실행은 루프를 작게 유지합니다. 직접적인 확장은 다음과 같습니다.
- 답하는 비용을 치르기 전에 후보를 스크리닝합니다. 제안된 질문 자체를 state로 삼아 그것에 대한 noul을 물으십시오. 원본 텍스트에서 답할 수 있는지, criteria 아래에서 한 가지를 뜻하는지, 대부분의 행에 적용되는지, 행마다 달라질지. 네 가지를 모두 충분한 신뢰도로 통과한 질문만 보냅니다.
- 상관된 특징을 쳐냅니다. dev 행에서 인코딩된 열 사이의 상관을 측정하고, 거의 중복인 것들을 군집화한 뒤, 각 군집에서 가장 명확하거나 가장 중요한 질문을 남깁니다.
- 간단한 기준선을 추가합니다. TF-IDF, 문자 수, 기타 구조적 특징을 단독으로 비교한 뒤, 발견된 열에 덧붙여 각각이 무엇을 기여하는지 측정합니다.
- 제안자 계열을 섞습니다. Anthropic, OpenAI, Google Gemini, 그리고 오픈소스 모델로 후보 배치를 생성한 뒤, TypeSafe에 도달하기 전에 병합하고 중복을 제거합니다. 서로 다른 계열은 한 제안자를 반복 호출하는 것보다 탐색을 더 넓혀야 합니다.
- 예측 모델과 방법을 비교합니다. 선형 또는 elastic-net 회귀, 서포트 벡터 회귀, 랜덤 포레스트, 그리고 하류 출력이 확률적일 때의 재캘리브레이션을 시도합니다. 발견된 특징이 CatBoost 밖에서도 도움이 되는지 확인합니다.
- 임베딩 기준선을 추가합니다. 임베딩은 질문 없이 노트를 수백 개의 숫자로 바꿉니다.
sentence-transformers/all-MiniLM-L6-v2는 로컬에서 실행되고, OpenAI의text-embedding-3-small은 호스팅 호출입니다. 발견된 열에 하나를 덧붙여, 그 열들이 담지 못한 무엇을 담고 있는지 측정합니다. - 검증을 배포에 맞춥니다. 미래를 예측할 때는 시간순 분할을, 관련된 행을 함께 유지해야 할 때는 그룹 분할을 사용하고, 최종 테스트 집합은 특징 발견과 모델 선택 모두에서 건드리지 않도록 남겨 둡니다.
- 정체 구간에서 멈춥니다. 교차 검증된 RMSE가 정해진 라운드 수 동안 개선되지 않거나, 질문이나 요청 예산에 도달하면 루프를 끝냅니다.
- 에이전트의 Goal 모드에서 더 긴 탐색을 실행합니다. 명시적인 지표, 예산, 정지 규칙을 주고, 더 많은 라운드를 제안·평가·개선하게 합니다.
- 안정성을 확인합니다. 시드나 데이터 조각을 바꿔 발견을 반복하고, 중요도가 한 번의 분할에 기대는 질문보다 계속 유용한 질문을 남깁니다.
playground에서 열기
이 공유 링크는 시음 노트 하나와 루프가 최종적으로 얻게 된 모든 질문을 담고 있습니다.
playground_link = make_playground_link(
NOTES[0], feature_questions(accepted), models=[TYPESAFE_MODEL]
)
display(
Markdown(
f"🔗 [Open the note + questions in the TypeSafe playground]({playground_link})"
)
)