SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2603.04855 · 2026

HACHIMI: Scalable and Controllable Student Persona Generation via Orchestrated Agents

Fei Tan, Aimin Zhou, Yilin Jiang, Xuanyu Yin, Jing Leng

arXiv · PDF · Open in the Atlas

Code that ran

5 of 16 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (3 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 1 violated it and 3 have no contract verdict; 1 failure of our instrument; 11 no recorded run), measured 2026-11-22; bodies inline only under a permissive licence as GitHub detects it. Not a reproduction of the paper's results.

"Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results. This record, dated, as JSON: /api/ran/2603.04855.json · badge: /api/ran/2603.04855.svg

Each repository's role is a label: a classifier's output recorded on the link, or, where no role is recorded, what the Papers with Code archive marked. The line under it names its basis. Neither says who wrote the code.

RepositoryRoleRan
ZeroLoss-Lab/HACHIMI community
Papers with Code archive
5 ran · 11 no recorded run
FunctionStatusWhere it lives
KeyClient Ran
ran on a synthesized input · constructed an object; no result was computed
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("fc9fc6697cd3248d")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, KeyClient: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import re
import random
import math
import os
import glob
import time
import hashlib
from typing import Any, Dict, List, Tuple, Optional
from collections import deque
import threading

ARGS = {
    'api_key': 'sk-test-1234567890abcdefghijklmnop',
    'base_url': 'https://api.example.com/v1',
    'qpm_init': 6,
    'capacity_max': 180,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
requests — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type KeyClient — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a KeyClient output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
RateLimiter Ran
ran on a synthesized input · constructed an object; no result was computed
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("d49640b34d90907b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, RateLimiter: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import time
import threading
from collections import deque

ARGS = {
    'qpm_init': 6,
    'capacity_max': 180,
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type RateLimiter — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a RateLimiter output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Whiteboard Ran
ran on a synthesized input · constructed an object; no result was computed
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("b965d9e0eb7eab5e")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned an instance of its own class, Whiteboard: the run constructed the object and did not call it, so no result was computed. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import torch
from copy import deepcopy
from typing import Any, Dict, List, Optional

# Build arguments for Whiteboard.__init__

# sid: int - a simple session ID
sid = 42

# sampling_hint: Optional[Dict[str, Any]] - can be None or a dict with sampling constraints
sampling_hint = {
    'temperature': 0.7,
    'top_k': 10,
    'max_tokens': 256
}

ARGS = {
    'sid': sid,
    'sampling_hint': sampling_hint
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type Whiteboard — the run constructed an instance of its own class and did not call it; no shape is stated — the instrument coerces any output to an array; for a Whiteboard output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
non_empty Ran
ran, violated the contract we drafted
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("75e3ce1507a92387")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned bool, ndim 0, dtype family float, shape () as a later run recorded it. The image it ran in was not recorded. If your copy returns the same type, ndim and shape on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import torch
import numpy as np

torch.manual_seed(0)

# Analyze the function: non_empty(v: Any) -> bool
# The function takes a single parameter 'v' of type Any
# and checks if it's non-empty based on various types:
# - None -> False
# - str -> True if stripped non-empty
# - list or dict -> True if len > 0
# - anything else -> True

# Create test values for each branch
test_none = None
test_empty_str = "   "
test_nonempty_str = "hello"
test_empty_list = []
test_nonempty_list = [1, 2, 3]
test_empty_dict = {}
test_nonempty_dict = {"key": "value"}
test_int = 42
test_tensor = torch.randn(2, 4)

# Define ARGS with the single parameter 'v'
# We'll test with a non-empty string value
ARGS = {
    "v": test_nonempty_str
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type bool, ndim 0, dtype family float, shape () — shape () as a later run recorded it
dtype family
int means an integer dtype; float means any other dtype the instrument saw, bool and complex included (fixture_check.py records only these two families)
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
try_json Ran
ran; our contract draft was wrong, not the code
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("c41f8df683acf35b")
How it ran

Ran in Syntology's sandbox on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API) and returned dict. The image it ran in was not recorded. If your copy returns the same type on these inputs, it matched what we observed; nothing here checks that. It does NOT mean the output is correct: nothing has compared it to the paper.

import json
import re
import random
import math
import os
import glob
import time
import hashlib
from typing import Any, Dict, List, Tuple, Optional

# Analyze the function: try_json(text: str) -> Dict[str, Any]
# The function takes a single parameter: text (str)
# It attempts to parse the text as JSON, with fallback logic

# Test cases for valid JSON parsing
ARGS = {
    "text": '{"key": "value", "number": 42, "nested": {"inner": true}}'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
none outside the standard library — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type dict — a dict output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a dict output the recorded ndim 0 and dtype family float describe that coercion, not the output, and are not stated
Environment
image not recorded; re-verify against syntology-harvest-sandbox:v5-deps; not recorded for this sample: the harness installs a cuda-to-cpu shim (cuda_shim.py) before every sample and no run records whether a sample needed it; seed: none by the harness; any seed is in the setup block
Orchestrator Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("ce5843002691a1b0")
How it ran

no invocation recorded.

agent_academic Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("6b306576c9d4dafa")
How it ran

no invocation recorded.

agent_health Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("ec4281b851c74c9e")
How it ran

no invocation recorded.

agent_scholar Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("8ed41240a82efe77")
How it ran

no invocation recorded.

agent_social_creative Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("b8ae30c84aaf133d")
How it ran

no invocation recorded.

agent_validator_deep Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("3a7663bfd41f138c")
How it ran

no invocation recorded.

agent_validator_fast Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("4ac37daff7ce1a18")
How it ran

no invocation recorded.

agent_values Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("222ef6c812fcca6f")
How it ran

no invocation recorded.

call_llm Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("abf2fd85dc0391da")
How it ran

no invocation recorded.

get_client_for_thread Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("1af4941fb3e2b087")
How it ran

no invocation recorded.

set_client_for_thread Not yet run
unverified (harvested, no recorded run)
ZeroLoss-Lab/HACHIMI/app_for_GPT4.1.py
code served (permissive licence) · get_code("a58d3e8d948e9b62")
How it ran

no invocation recorded.

Also present in other papers' code

Of 16 samples lifted from this paper's linked repositories, 16 carry a function identity and 0 do not. Of the 16 distinct functions with an identity, 0 are also present in other papers whose code we hold (0 with at least one paper dated earlier, 0 the same month, 0 later) and 16 are only here among the 52,977 papers whose code Syntology holds (as of 2026-09-26).

Only here means: among the 52,977 papers whose code Syntology holds (51,887 with lifted samples, 1,090 through their assemblies only), as of 2026-09-26, no other paper's lifted samples or current assembly bricks contain an identical function body after normalisation.

Presence, not direction: an identical function body in two papers' repositories does not say who wrote it, which came from which, or that either paper's method rests on it. Matching is exact after normalisation, so a renamed or edited function is not matched. Dates are when papers appeared, not when code was written. Only .py files the harvester lifted are covered; notebooks are not read. A new-style arXiv id YYMM.NNNN dates a paper to the month 20YY-MM; otherwise Paper.year dates it to the year; otherwise it is undated. Dates are when the paper appeared, not when the code was written. Read 2026-11-22.

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Student Personas (SPs) are emerging as infrastructure for educational LLMs, yet prior work often relies on ad-hoc prompting or hand-crafted profiles with limited control over educational theory and population distributions. We formalize this as Theory-Aligned and Distribution-Controllable Persona Generation (TAD-PG) and introduce HACHIMI, a multi-agent Propose-Validate-Revise framework that generates theory-aligned, quotacontrolled personas. HACHIMI factorizes each persona into a theory-anchored educational schema, enforces developmental and psychological constraints via a neuro-symbolic validator, and combines stratified sampling with semantic deduplication to reduce mode collapse. The resulting HACHIMI-1M corpus comprises 1 million personas for Grades 1-12. Intrinsic evaluation shows near-perfect schema validity, accurate quotas, and substantial diversity, while external evaluation instantiates personas as student agents answering CEPS and PISA 2022 surveys; across 16 cohorts, math and curiosity/growth constructs align strongly between humans and agents, whereas classroom-climate and well-being constructs are only moderately aligned, revealing a fidelity gradient. All personas are generated with Qwen2.5-72B, and HACHIMI provides a standardized synthetic student population for group-level benchmarking and social-science simulations. Resources available at https: //github.com/ZeroLoss-Lab/HACHIMI.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2603.04855")
get_code_for_paper("2603.04855")
have("2603.04855")

The run record, dated, one paper per request, free:

curl https://syntology.ai/api/ran/2603.04855.json

A badge for a README (the split and the date, never a ratio):

[![Syntology run record](https://syntology.ai/api/ran/2603.04855.svg)](https://syntology.ai/paper/2603.04855)

Connect an agent — have() is free.