SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2004.08449 · ACL · 2020

Can You Put it All Together: Evaluating Conversational Agents' Ability to Blend Skills

Mary Williamson, Jason Weston, Eric Smith, Kurt Shuster, Y-Lan Boureau

arXiv · PDF · Open in the Atlas

Code that ran

2 of 4 samples (functions and classes) lifted from this paper's linked repositories executed on a synthesized input in Syntology's jail (0 of them constructed an object rather than computing a result; 0 honoured the contract we drafted, 0 violated it and 2 have no contract verdict; 0 failures of our instrument; 2 no recorded run), measured 2026-11-22; repository links from the Papers with Code archive attributed under CC BY-SA, no affiliation; 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/2004.08449.json · badge: /api/ran/2004.08449.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
devJWSong/gpt2-chatbot-pytorch community (classifier label)
classifier basis not recorded on the link
2 ran · 2 no recorded run
FunctionStatusWhere it lives
load_persona Ran
ran on a synthesized input
devJWSong/gpt2-chatbot-pytorch/src/process_data.py
code served (permissive licence) · get_code("29cce3c0c0462dca")
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 tuple. 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 torch
from unittest.mock import Mock, patch
import json

class MockTokenizer:
    """Stand-in tokenizer that mimics HuggingFace tokenizer interface"""
    def tokenize(self, text):
        # Simple tokenization: split on spaces and add space prefix
        words = text.split()
        tokens = []
        for i, word in enumerate(words):
            if i == 0:
                tokens.append(word)
            else:
                tokens.append('Ġ' + word)
        return tokens
    
    def convert_tokens_to_string(self, tokens):
        # Convert tokens back to string
        result = []
        for token in tokens:
            if token.startswith('Ġ'):
                result.append(token[1:])
            else:
                result.append(token)
        return ' '.join(result)

def mock_urlopen(url):
    """Mock urlopen to return fake persona chat data"""
    mock_response = Mock()
    
    # Create minimal persona chat dataset structure
    fake_data = {
        'train': [
            {
                'utterances': [
                    {
                        'history': ['Hello world .', 'How are you ?']
                    }
                ]
            },
            {
                'utterances': [
                    {
                        'history': ['I am fine', 'That is great']
                    }
                ]
            }
        ],
        'valid': [
            {
                'utterances': [
                    {
                        'history': ['Good morning', 'Have a nice day']
                    }
                ]
            }
        ]
    }
    
    json_str = json.dumps(fake_data)
    mock_response.read.return_value = json_str.encode('utf-8')
    mock_response.__enter__ = Mock(return_value=mock_response)
    mock_response.__exit__ = Mock(return_value=None)
    
    return mock_response

# Patch urllib before defining ARGS
import urllib.request
original_urlopen = urllib.request.urlopen
urllib.request.urlopen = mock_urlopen

tokenizer = MockTokenizer()
train_frac = 0.6

ARGS = {
    'tokenizer': tokenizer,
    'train_frac': train_frac
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
torch, tqdm — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type tuple — a tuple output has no array shape of its own; none is 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
merge_data Ran
ran on a synthesized input
devJWSong/gpt2-chatbot-pytorch/src/load_data.py
code served (permissive licence) · get_code("0a36aa40cb307a40")
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 tuple. 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 torch
import numpy as np
from unittest.mock import Mock

torch.manual_seed(0)

# Create a minimal tokenizer stand-in
class MinimalTokenizer:
    def __init__(self):
        self.vocab_size = 1000
    
    def encode(self, text):
        return [1, 2, 3]
    
    def decode(self, ids):
        return "decoded text"

# Create a minimal args object
class Args:
    def __init__(self):
        self.train_frac = 0.8

# Mock the load functions to return valid dialogue data
def mock_load_daily(tokenizer, train_frac):
    train_dialogues = [
        [{"text": "hello", "tokens": [1, 2]}, {"text": "hi", "tokens": [3, 4]}],
        [{"text": "how are you", "tokens": [5, 6]}, {"text": "fine", "tokens": [7, 8]}]
    ]
    valid_dialogues = [
        [{"text": "goodbye", "tokens": [9, 10]}]
    ]
    num_train = 4
    num_valid = 1
    return train_dialogues, valid_dialogues, num_train, num_valid

def mock_load_empathetic(tokenizer, train_frac):
    train_dialogues = [
        [{"text": "I'm sad", "tokens": [11, 12]}, {"text": "I understand", "tokens": [13, 14]}]
    ]
    valid_dialogues = [
        [{"text": "cheer up", "tokens": [15, 16]}]
    ]
    num_train = 2
    num_valid = 1
    return train_dialogues, valid_dialogues, num_train, num_valid

def mock_load_persona(tokenizer, train_frac):
    train_dialogues = [
        [{"text": "I like dogs", "tokens": [17, 18]}, {"text": "me too", "tokens": [19, 20]}]
    ]
    valid_dialogues = [
        [{"text": "dogs are great", "tokens": [21, 22]}]
    ]
    num_train = 2
    num_valid = 1
    return train_dialogues, valid_dialogues, num_train, num_valid

def mock_load_blended(tokenizer, train_frac):
    train_dialogues = [
        [{"text": "hello there", "tokens": [23, 24]}]
    ]
    valid_dialogues = [
        [{"text": "hi again", "tokens": [25, 26]}]
    ]
    num_train = 1
    num_valid = 1
    return train_dialogues, valid_dialogues, num_train, num_valid

# Patch the load functions in the module where merge_data is defined
import sys
current_module = sys.modules[__name__]
current_module.load_daily = mock_load_daily
current_module.load_empathetic = mock_load_empathetic
current_module.load_persona = mock_load_persona
current_module.load_blended = mock_load_blended

ARGS = {
    'tokenizer': MinimalTokenizer(),
    'args': Args()
}
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 tuple — a tuple output has no array shape of its own; none is 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
load_daily Not yet run
unverified (harvested, no recorded run)
devJWSong/gpt2-chatbot-pytorch/src/process_data.py
code served (permissive licence) · get_code("6b86de55711c20fe")
How it ran

no invocation recorded.

load_empathetic Not yet run
unverified (harvested, no recorded run)
devJWSong/gpt2-chatbot-pytorch/src/process_data.py
code served (permissive licence) · get_code("63710ca3717d7d4c")
How it ran

no invocation recorded.

Also present in other papers' code

Of 4 samples lifted from this paper's linked repositories, 4 carry a function identity and 0 do not. Of the 4 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 4 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

Being engaging, knowledgeable, and empathetic are all desirable general qualities in a conversational agent. Previous work has introduced tasks and datasets that aim to help agents to learn those qualities in isolation and gauge how well they can express them. But rather than being specialized in one single quality, a good open-domain conversational agent should be able to seamlessly blend them all into one cohesive conversational flow. In this work, we investigate several ways to combine models trained towards isolated capabilities, ranging from simple model aggregation schemes that require minimal additional training, to various forms of multi-task training that encompass several skills at all training stages. We further propose a new dataset, Blended-SkillTalk, to analyze how these capabilities would mesh together in a natural conversation, and compare the performance of different architectures and training schemes. Our experiments show that multi-tasking over several tasks that focus on particular capabilities results in better blended conversation performance compared to models trained on a single skill, and that both unified or two-stage approaches perform well if they are constructed to avoid unwanted bias in skill selection or are fine-tuned on our new task.

For agents

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

get_harvested_code_for_paper("2004.08449")
get_code_for_paper("2004.08449")
have("2004.08449")

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

curl https://syntology.ai/api/ran/2004.08449.json

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

[![Syntology run record](https://syntology.ai/api/ran/2004.08449.svg)](https://syntology.ai/paper/2004.08449)

Connect an agent — have() is free.