SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2601.16449 · 2026

Emotion-LLaMAv2 and MMEVerse: A New Framework and Benchmark for Multimodal Emotion Understanding

Kai Wang, Zhi-Qi Cheng, Jun-Yan He, Zebang Cheng, Yuxiang Lin, Xiaojiang Peng, Yifei Dong, Fengyi Wu, Qiyu Hu, Jingyi Chen, Shuyuan Tu, Bao Peng, and 1 more

arXiv · PDF · Open in the Atlas

Code that ran

9 of 10 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, 1 violated it and 0 have no contract verdict; 8 failures of our instrument; 1 no recorded run); 1 of them was also run on the host, not in the jail, by the peer-agreement lane that set its level, 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/2601.16449.json · badge: /api/ran/2601.16449.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
ooochen-30/Emotion-LLaMA-v2 canonical (classifier label)
classifier basis: census
9 ran · 1 no recorded run
FunctionStatusWhere it lives
list_of_str Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/eval_mme_emotion_llama_v2.py
code served (permissive licence) · get_code("747d1d6a93d4f89d")
How it ran

Ran in Syntology's jail on this synthesized input (drafted by a language model, Claude Haiku 4.5, prompt python_fixture_v1) and returned list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. The comparison behind this sample's level is a different run: the peer-agreement lane ran it again on the host, not in the jail (report dated 2026-09-05T00:25:35Z). If your copy returns the same type and ndim 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)

# The function list_of_str takes a single argument 'arg'
# It calls arg.split(',') which means arg must be a string
# Then it maps str over the result and converts to list

ARGS = {
    'arg': 'hello,world,test'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
numpy, torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
disabled_train Ran
ran, violated the contract we drafted
ooochen-30/Emotion-LLaMA-v2/minigpt4/models/base_model.py
code served (permissive licence) · get_code("4cb732f513d69dfd")
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) and returned SimpleModel. 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 torch.nn as nn

torch.manual_seed(0)

class SimpleModel(nn.Module):
    def __init__(self):
        super().__init__()
        self.linear = nn.Linear(4, 3)
    
    def forward(self, x):
        return self.linear(x)

model = SimpleModel()

ARGS = {
    "self": model,
    "mode": True
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type SimpleModel — a SimpleModel output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a SimpleModel 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
escape_markdown Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/app_temp.py
code served (permissive licence) · get_code("b495489d235db0b8")
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) and returned str. 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

ARGS = {
    'text': '<hello> world <test>'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1
Imports
torch — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type str — a str output has no array shape of its own; none is stated — the instrument coerces any output to an array; for a str 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
func_read_key_from_csv Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_hit_ov.py
code served (permissive licence) · get_code("2cd2ea6202a00976")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 pandas as pd
import tempfile
import os

# Create a temporary CSV file with test data
temp_dir = tempfile.mkdtemp()
csv_path = os.path.join(temp_dir, "test.csv")

# Write test CSV data
df = pd.DataFrame({
    'name': ['Alice', 'Bob', 'Charlie'],
    'age': [25, 30, None],
    'city': ['NYC', None, 'LA']
})
df.to_csv(csv_path, index=False)

ARGS = {
    'csv_path': csv_path,
    'key': 'age'
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
load_jsonl Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_split.py
code served (permissive licence) · get_code("a118c735eadfbee8")
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, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 tempfile
import os

# Create a temporary JSONL file with test data
temp_dir = tempfile.mkdtemp()
temp_file = os.path.join(temp_dir, 'test.jsonl')

# Write sample JSONL data
with open(temp_file, 'w', encoding='utf-8') as f:
    f.write(json.dumps({'mode': 'train', 'answer': 'answer1', 'new_response': '  Response One  '}) + '\n')
    f.write(json.dumps({'mode': 'test', 'answer': 'answer2', 'response': '  Response Two  '}) + '\n')
    f.write(json.dumps({'mode': 'train', 'answer': 'answer3', 'new_response': '  Response Three  '}) + '\n')
    f.write(json.dumps({'answer': 'answer4', 'response': '  Response Four  '}) + '\n')

ARGS = {
    'file_path': temp_file,
    'split': 'train'
}
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 tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
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
load_jsonl Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_split_sentiment_no_neutral.py
code served (permissive licence) · get_code("af8fdb6ca114a86c")
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, ndim 2, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 tempfile
import os

# Create a temporary JSONL file with test data
temp_dir = tempfile.mkdtemp()
file_path = os.path.join(temp_dir, "test.jsonl")

# Write test data to the JSONL file
test_data = [
    {"mode": "train", "answer": "YES", "response": "This is a response"},
    {"mode": "train", "answer": "NO", "new_response": "This is a new response"},
    {"mode": "train", "answer": "NEUTRAL", "response": "This should be skipped"},
    {"mode": "test", "answer": "YES", "response": "Test response"},
]

with open(file_path, 'w', encoding='utf-8') as f:
    for item in test_data:
        f.write(json.dumps(item) + '\n')

ARGS = {
    "file_path": file_path,
    "split": "train"
}
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 tuple, ndim 2, dtype family float — a tuple output has no array shape of its own; none is stated
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
load_multilabel_data Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_multilabel.py
code served (permissive licence) · get_code("324d127359c81610")
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, ndim 3, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 tempfile
import os

def split_labels(text):
    return [label.strip().lower() for label in re.split('[;,]', text) if label.strip()]

def load_multilabel_data(jsonl_path):
    true_labels = []
    pred_labels = []
    with open(jsonl_path, 'r', encoding='utf-8') as f:
        for line in f:
            data = json.loads(line.strip())
            true = split_labels(data['answer'])
            pred = split_labels(data['response'])
            true_labels.append(true)
            pred_labels.append(pred)
    return (true_labels, pred_labels)

# Create a temporary JSONL file with sample multilabel data
temp_dir = tempfile.mkdtemp()
jsonl_file = os.path.join(temp_dir, "test_data.jsonl")

# Write sample data to JSONL file
sample_data = [
    {"answer": "cat, dog; animal", "response": "cat; pet, dog"},
    {"answer": "red; blue, green", "response": "red, blue; color"},
]

with open(jsonl_file, 'w', encoding='utf-8') as f:
    for item in sample_data:
        f.write(json.dumps(item) + '\n')

ARGS = {
    "jsonl_path": jsonl_file,
}
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 tuple, ndim 3, dtype family float — a tuple output has no array shape of its own; none is stated
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
split_labels Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_multilabel.py
code served (permissive licence) · get_code("c90df7b03ec16987")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 re

def split_labels(text):
    return [label.strip().lower() for label in re.split('[;,]', text) if label.strip()]

# The function takes a single parameter 'text' which is a string
# It splits by semicolons or commas, strips whitespace, converts to lowercase, and filters empty strings

ARGS = {
    "text": "Label1; Label2, LABEL3 ; label4"
}
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 list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
string_to_list Ran
ran; our contract draft was wrong, not the code
ooochen-30/Emotion-LLaMA-v2/evaluation/score_hit_ov.py
code served (permissive licence) · get_code("9e68bdcd13c06ee1")
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 list, ndim 1, dtype family float; no output shape is stated. The image it ran in was not recorded. If your copy returns the same type and ndim 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 re
import numpy as np
import pandas as pd

# Test case 1: string with brackets and quoted items
test_str_1 = "['apple', 'banana', 'cherry']"

# Test case 2: empty string
test_str_2 = ''

# Test case 3: string with commas but no brackets
test_str_3 = "item1, item2, item3"

# Test case 4: list input (already a list)
test_list = ['a', 'b', 'c']

# Test case 5: numpy array
test_array = np.array(['x', 'y', 'z'])

# Test case 6: NaN value
test_nan = pd.NA

# Test case 7: string with double quotes
test_str_7 = '["one", "two", "three"]'

# Using the first test case as the main argument
ARGS = {
    'str': test_str_1
}
Setup block drafted by
a language model, Claude Haiku 4.5, prompt python_fixture_v1, through the batch API
Imports
numpy, pandas — import names as written, not package names to install; a name the repository itself provides is not a package
Returned
type list, ndim 1, dtype family float — a list output has no array shape of its own; none is stated
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
read_wheel_to_map Not yet run
unverified (harvested, no recorded run)
ooochen-30/Emotion-LLaMA-v2/evaluation/score_hit_ov.py
code served (permissive licence) · get_code("cce73f408489fe3d")
How it ran

no invocation recorded.

Also present in other papers' code

Of 10 samples lifted from this paper's linked repositories, 10 carry a function identity and 0 do not. Of the 10 distinct functions with an identity, 3 are also present in other papers whose code we hold (3 with at least one paper dated earlier, 0 the same month, 1 later) and 7 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.

FunctionAlso present inEarliest-datedOthers named
disabled_train 173 other papers
167 dated earlier, 6 later · 5 only as an assembly brick
2012.09841 (2020-12)
Taming Transformers for High-Resolution Image Synthesis
2102.05379 (2021-02)
Argmax Flows and Multinomial Diffusion: Learning Categorical Distributions
2104.07652 (2021-04)
Geometry-Free View Synthesis: Transformers and no 3D Priors
and 170 more not named here
list_of_str 4 other papers
4 dated earlier
2312.16862 (2023-12)
TinyGPT-V: Efficient Multimodal Large Language Model via Small Backbones
2404.00511 (2024-04)
MIPS at SemEval-2024 Task 3: Multimodal Emotion-Cause Pair Extraction in Conversations with Multimodal Language Models
2405.09981 (2024-05)
Published at ICLR 2024 Workshop on Reliable and Responsible Foundation Models ADVERSARIAL ROBUSTNESS FOR VISUAL GROUND-ING OF MULTIMODAL LARGE LANGUAGE MODELS
and 1 more not named here
escape_markdown 2 other papers
2 dated earlier
2402.02544 (2024-02)
LHRS-Bot: Empowering Remote Sensing with VGI-Enhanced Large Multimodal Language Model
2411.09301 (2024-11)
LHRS-Bot-Nova: Improved Multimodal Large Language Model for Remote Sensing Vision-Language Interpretation

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

Understanding human emotions from multimodal signals poses a significant challenge in affective computing and humanrobot interaction. While multimodal large language models (MLLMs) have excelled in general vision-language tasks, their capabilities in emotional reasoning remain limited. The field currently suffers from a scarcity of large-scale datasets with high-quality, descriptive emotion annotations and lacks standardized benchmarks for evaluation. Our preliminary framework, Emotion-LLaMA, pioneered instruction-tuned multimodal learning for emotion reasoning but was restricted by explicit face detectors, implicit fusion strategies, and low-quality training data with limited scale. To address these limitations, we present Emotion-LLaMAv2 and the MMEVerse benchmark, establishing an end-to-end pipeline together with a standardized evaluation setting for emotion recognition and reasoning. Emotion-LLaMAv2 introduces three key advances. First, an end-to-end multiview encoder eliminates external face detection and captures nuanced emotional cues via richer spatial and temporal multiview tokens. Second, a Conv Attention pre-fusion module is designed to enable simultaneous local and global multimodal feature interactions external to the LLM backbone. Third, a perception-to-cognition curriculum instruction tuning scheme within the LLaMA2 backbone unifies emotion recognition and free-form emotion reasoning. To support large-scale training and reproducible evaluation, MMEVerse aggregates twelve publicly available emotion datasets, including IEMOCAP, MELD, DFEW, and MAFW, into a unified multimodal instruction format. The data are re-annotated via a multi-agent pipeline involving Qwen2 Audio, Qwen2.5 VL, and GPT 4o, producing 130k training clips and 36k testing clips across 18 evaluation benchmarks. Experiments demonstrate that Emotion-LLaMAv2 consistently outperforms representative open source MLLMs such as Qwen2.5 Omni and AffectGPT, achieving state-of-the-art performance on MER-UniBench and MMEVerse-Bench, with improved generalization and more structured multimodal reasoning behavior. Overall, Emotion-LLaMAv2 and MMEVerse provide a reproducible research foundation for studying multimodal emotion recognition and reasoning, and represent a step toward connecting low-level multimodal perception with emotion-centered language reasoning. Our code and benchmark are available at https://github.com/ooochen-30/Emotion-LLaMA-v2.

For agents

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

get_harvested_code_for_paper("2601.16449")
get_code_for_paper("2601.16449")
have("2601.16449")

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

curl https://syntology.ai/api/ran/2601.16449.json

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

[![Syntology run record](https://syntology.ai/api/ran/2601.16449.svg)](https://syntology.ai/paper/2601.16449)

Connect an agent — have() is free.