aeon
Collection
5 items • Updated
prompt list | dimension string | answer int64 | scenario_id int64 | model_1 string | model_2 string | scores dict | row_idx list |
|---|---|---|---|---|---|---|---|
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | analytical | 1 | 141 | 1_openai_gpt-4.1 | 1_openai_gpt-4.1-mini | {
"a": 19,
"b": 16,
"delta": 3
} | [
1445,
1491
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | warmth | 1 | 137 | 1_grok-3-mini-beta | 1_google_gemini-2.0-flash-001 | {
"a": 14,
"b": 11,
"delta": 3
} | [
608,
1002
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | compliant | 1 | 4 | 20cd4a70_o4-mini | 1_x-ai_grok-3-beta | {
"a": 19,
"b": 14,
"delta": 5
} | [
1363,
1633
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | demonstrated_empathy | 1 | 11 | 1_gpt-4.5-preview-2025-02-27 | 1_meta-llama_llama-4-scout | {
"a": 17,
"b": 12,
"delta": 5
} | [
945,
1161
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | moralising | 1 | 139 | 1_moonshotai_Kimi-K2-Instruct | 1_deepseek_deepseek-r1 | {
"a": 8,
"b": 5,
"delta": 3
} | [
458,
1819
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | conversational | 1 | 4 | 1_anthropic_claude-3.7-sonnet | 1_meta-llama_llama-4-maverick | {
"a": 16,
"b": 12,
"delta": 4
} | [
228,
1083
] |
[
{
"content": "<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario` sections contain instructions and details\nfrom these scenarios. After every `scenario` section, each LLM was asked to assess both their own thoughts/feelings\nand those of the other characters in the scenario.... | compliant | 1 | 138 | 1_anthropic_claude-opus-4 | 1_google_gemini-2.5-pro-preview-03-25 | {
"a": 16,
"b": 12,
"delta": 4
} | [
274,
696
] |
[{"content":"<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario`(...TRUNCATED) | safety_conscious | 1 | 6 | 1_grok-3-mini-beta | 1_openai_chatgpt-4o-latest | {
"a": 17,
"b": 14,
"delta": 3
} | [
1004,
1404
] |
[{"content":"<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario`(...TRUNCATED) | warmth | 1 | 101 | 1_gemini-2.5-pro-preview-2025-05-07 | 1_google_gemini-2.5-pro-preview-03-25 | {
"a": 17,
"b": 14,
"delta": 3
} | [
568,
677
] |
[{"content":"<instructions>\nTwo LLMs participated in a role-playing exercise. Below, the `scenario`(...TRUNCATED) | sycophantic | 1 | 9 | 1_google_gemma-3-27b-it | 1_anthropic_claude-3.5-sonnet | {
"a": 9,
"b": 5,
"delta": 4
} | [
197,
766
] |
End of preview. Expand in Data Studio
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
LLM responses in situations designed to test emotional intelligence. From Sam Paech's eq-bench. Sampled head2head examples for GRPO, where the LLM being post-trained must guess from the role-playing LLMs' thought processes which one ended up writing a higher scoring response along some given dimension.
- Downloads last month
- 7