queelius/rlm-qwen3-4b-sft-005
Updated • 13
oolong_row_id int64 10M 113M | context_len int64 1.02k 8.19k | task stringclasses 6
values | turn int64 0 2 | completion stringclasses 33
values | variant unknown | correct bool 2
classes | teaching_set stringclasses 2
values |
|---|---|---|---|---|---|---|---|
111,010,042 | 2,048 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,042 | 2,048 | TASK_TYPE.MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,042 | 2,048 | TASK_TYPE.MOST_FREQ | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"which user has the most instances with the label (\w[\w ]*?)\?", question).group(1).strip()
users = Counter(r["user"] for r in sub if r["label"] == lab)
FINAL_TEXT("User: " + users.most_common(1)[0][0])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,018 | 1,024 | TASK_TYPE.NUMERIC_ONE_CLASS | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
10,000,018 | 1,024 | TASK_TYPE.NUMERIC_ONE_CLASS | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
10,000,018 | 1,024 | TASK_TYPE.NUMERIC_ONE_CLASS | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"classified as label '([^']+)'", question).group(1)
FINAL_TEXT("Answer: " + str(sum(1 for r in sub if r["label"] == lab)))
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,041 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,041 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,041 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
lab, u1, u2 = re.search(r"which user has more instances with the label (\w[\w ]*?): User (\d+) or User (\d+)\?", question).groups()
c1 = sum(1 for r in sub if r["user"] == u1 and r["label"] == lab.strip())
c2 = sum(1 for r in sub if r["user"] == u2 and r["label"] == ... | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
110,010,017 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
110,010,017 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
110,010,017 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
lab, d = re.search(r"was label '([^']+)' more common, less common, or the same frequency before (\d{4}-\d{2}-\d{2})", question).groups()
d = datetime.strptime(d, "%Y-%m-%d").date()
before = sum(1 for r in sub if r["label"] == lab and r["date"] < d) / (sum(1 for r in ... | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,004 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
10,000,004 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
10,000,004 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
12,000,033 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
12,000,033 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
12,000,033 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,013 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,013 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,013 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
11,000,029 | 2,048 | TASK_TYPE.LEAST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,029 | 2,048 | TASK_TYPE.LEAST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,029 | 2,048 | TASK_TYPE.LEAST_FREQ | 2 | ```python
from collections import Counter
sub = rows
counts = Counter(r["label"] for r in sub if r["label"])
present = [l for l in labels if counts[l] > 0] or labels
FINAL_TEXT("Label: " + min(present, key=lambda l: counts[l]))
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
12,000,044 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
12,000,044 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
12,000,044 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"classified as label '([^']+)'", question).group(1)
FINAL_TEXT("Answer: " + str(sum(1 for r in sub if r["label"] == lab)))
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | false | teach-oolong-004 |
110,010,027 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
110,010,027 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
110,010,027 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"label '([^']+)' occur more frequently than the label '([^']+)'", question).groups()
by_month = Counter((r["date"].year, r["date"].month, r["label"]) for r in sub if r["label"] in (a, b))
months = {(y, m) for (y, m, _) in by_month}
FINAL_TEXT("Answe... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
111,010,047 | 2,048 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
111,010,047 | 2,048 | TASK_TYPE.MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
111,010,047 | 2,048 | TASK_TYPE.MOST_FREQ | 2 | ```python
from collections import Counter
MONTHS = ["january", "february", "march", "april", "may", "june", "july", "august", "september", "october", "november", "december"]
month = MONTHS.index(re.search(r"occur in (\w+) of any year", question).group(1).lower()) + 1
sub = [r for r in rows if r["date"].month == month]
... | {
"chunk": 24,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
111,010,044 | 2,048 | TASK_TYPE.REPRESENTED_N_TIMES | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,044 | 2,048 | TASK_TYPE.REPRESENTED_N_TIMES | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,044 | 2,048 | TASK_TYPE.REPRESENTED_N_TIMES | 2 | ```python
from collections import Counter
sub = rows
k = int(re.search(r"represented exactly (\d+) times", question).group(1))
FINAL_TEXT("Answer: " + str(sum(1 for c in Counter(r["date"] for r in sub).values() if c == k)))
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
112,010,047 | 4,096 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
112,010,047 | 4,096 | TASK_TYPE.MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
112,010,047 | 4,096 | TASK_TYPE.MOST_FREQ | 2 | ```python
from collections import Counter
MONTHS = ["january", "february", "march", "april", "may", "june", "july", "august", "september", "october", "november", "december"]
month = MONTHS.index(re.search(r"occur in (\w+) of any year", question).group(1).lower()) + 1
sub = [r for r in rows if r["date"].month == month]
... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
13,000,051 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
13,000,051 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
13,000,051 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"classified as label '([^']+)'", question).group(1)
FINAL_TEXT("Answer: " + str(sum(1 for r in sub if r["label"] == lab)))
``` | {
"chunk": 12,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
110,010,030 | 1,024 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
110,010,030 | 1,024 | TASK_TYPE.MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
110,010,030 | 1,024 | TASK_TYPE.MOST_FREQ | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"label '([^']+)' the single most frequently occur?ring label", question).group(1)
by_month = {}
for r in sub:
if r["label"]:
by_month.setdefault((r["date"].year, r["date"].month), Counter())[r["label"]] += 1
wins = sum(1 for c in by_month... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
113,010,053 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
113,010,053 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
113,010,053 | 8,192 | TASK_TYPE.NUMERIC_ONE_CLASS | 2 | ```python
from collections import Counter
MONTHS = ["january", "february", "march", "april", "may", "june", "july", "august", "september", "october", "november", "december"]
month = MONTHS.index(re.search(r"occur in (\w+) of any year", question).group(1).lower()) + 1
sub = [r for r in rows if r["date"].month == month]
... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
11,000,040 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,040 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,040 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
12,000,045 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
12,000,045 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
12,000,045 | 4,096 | TASK_TYPE.NUMERIC_ONE_CLASS | 2 | ```python
from collections import Counter
sub = rows
lab = re.search(r"classified as label '([^']+)'", question).group(1)
FINAL_TEXT("Answer: " + str(sum(1 for r in sub if r["label"] == lab)))
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | false | teach-oolong-004 |
12,000,037 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
12,000,037 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
12,000,037 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,011 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 24,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
10,000,011 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 24,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
10,000,011 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 24,
"prompt": "b",
"filter": "person"
} | false | teach-oolong-004 |
11,000,032 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,032 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,032 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
12,000,006 | 4,096 | TASK_TYPE.LEAST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
12,000,006 | 4,096 | TASK_TYPE.LEAST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
12,000,006 | 4,096 | TASK_TYPE.LEAST_FREQ | 2 | ```python
from collections import Counter
sub = rows
counts = Counter(r["label"] for r in sub if r["label"])
present = [l for l in labels if counts[l] > 0] or labels
FINAL_TEXT("Label: " + min(present, key=lambda l: counts[l]))
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
13,000,057 | 8,192 | TASK_TYPE.SECOND_MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
13,000,057 | 8,192 | TASK_TYPE.SECOND_MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
13,000,057 | 8,192 | TASK_TYPE.SECOND_MOST_FREQ | 2 | ```python
from collections import Counter
sub = rows
rank = 1 if "second most" in question else 0
FINAL_TEXT("User: " + Counter(r["user"] for r in sub).most_common()[rank][0])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,030 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,030 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,030 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 12,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
112,010,041 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
112,010,041 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
112,010,041 | 4,096 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
lab, u1, u2 = re.search(r"which user has more instances with the label (\w[\w ]*?): User (\d+) or User (\d+)\?", question).groups()
c1 = sum(1 for r in sub if r["user"] == u1 and r["label"] == lab.strip())
c2 = sum(1 for r in sub if r["user"] == u2 and r["label"] == ... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,012 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,012 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
10,000,012 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
11,000,035 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 24,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,035 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 24,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
11,000,035 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 24,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
113,010,046 | 8,192 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
113,010,046 | 8,192 | TASK_TYPE.MOST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
113,010,046 | 8,192 | TASK_TYPE.MOST_FREQ | 2 | ```python
from collections import Counter
sub = rows
rank = 1 if "second most" in question else 0
d = Counter(r["date"] for r in sub).most_common()[rank][0]
FINAL_TEXT(f"Date: {d.month:02d}/{d.day:02d}/{d.year:04d}")
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
111,010,037 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,037 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
111,010,037 | 2,048 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
ids = set(re.findall(r"\d+", re.search(r"associated with user IDs? ([\d, ]+)\.", question).group(1)))
sub = [r for r in rows if r["user"] in ids]
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
113,010,047 | 8,192 | TASK_TYPE.REPRESENTED_N_TIMES | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
113,010,047 | 8,192 | TASK_TYPE.REPRESENTED_N_TIMES | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
113,010,047 | 8,192 | TASK_TYPE.REPRESENTED_N_TIMES | 2 | ```python
from collections import Counter
sub = rows
k = int(re.search(r"represented exactly (\d+) times", question).group(1))
FINAL_TEXT("Answer: " + str(sum(1 for c in Counter(r["date"] for r in sub).values() if c == k)))
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person"
} | true | teach-oolong-004 |
110,010,022 | 1,024 | TASK_TYPE.LEAST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
110,010,022 | 1,024 | TASK_TYPE.LEAST_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
110,010,022 | 1,024 | TASK_TYPE.LEAST_FREQ | 2 | ```python
from collections import Counter
lo, hi = (datetime.strptime(s, "%b %d, %Y").date() for s in re.search(r"occur between (\w+ \d+, \d{4}) and (\w+ \d+, \d{4}), inclusive", question).groups())
sub = [r for r in rows if lo <= r["date"] <= hi]
counts = Counter(r["label"] for r in sub if r["label"])
present = [l for... | {
"chunk": 12,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,005 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,005 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
10,000,005 | 1,024 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 6,
"prompt": "b",
"filter": "person"
} | true | teach-oolong-004 |
13,000,043 | 8,192 | TASK_TYPE.RELATIVE_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
13,000,043 | 8,192 | TASK_TYPE.RELATIVE_FREQ | 1 | ```python
import re, json
from datetime import datetime
text = request["input"]
rows = []
for line in text.split("\n"):
m = re.match(r"Date: (.*?) \|\| User: (\d+) \|\| Instance: (.*)$", line)
if m:
rows.append({"date": datetime.strptime(m.group(1).strip(), "%b %d, %Y").date(), "user": m.group(2), "text... | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
13,000,043 | 8,192 | TASK_TYPE.RELATIVE_FREQ | 2 | ```python
from collections import Counter
sub = rows
a, b = re.search(r"is label '([^']+)' more common, less common, or the same frequency as label '([^']+)'", question).groups()
counts = Counter(r["label"] for r in sub if r["label"])
x = "more common than" if counts[a] > counts[b] else "less common than" if counts[a] ... | {
"chunk": 6,
"prompt": "a",
"filter": "person_item"
} | true | teach-oolong-004 |
10,000,023 | 1,024 | TASK_TYPE.MOST_FREQ | 0 | ```python
text = request["input"]
print(len(text), "chars")
print(text[:500])
print("...")
print(text[-600:])
``` | {
"chunk": 6,
"prompt": "b",
"filter": "person_item"
} | true | teach-oolong-004 |
End of preview. Expand in Data Studio
rlm-teaching-oolong-cells
Controller cells taught on OOLONG-synth rows, by row id, without the benchmark text.
Our controller cells (1174 rows) from scripted-expert trajectories on OOLONG-synth validation
rows up to 8k tokens (even context windows). The benchmark text is not included because the
source dataset (oolongbench/oolong-synth, revision f0d59eaf) declares no license; each row
carries oolong_row_id (the source id), context_len, task, turn, completion (our cell),
variant, correct, teaching_set. To rebuild the full chat contexts, rerun the expert in
https://github.com/queelius/rlm (experiments/selective_recursion_20261019/teach_oolong.py)
against the source dataset at that revision.
Citation: Towell, A. (2026). RLM research repository, https://github.com/queelius/rlm.
- Downloads last month
- 20