Skip to content

[OpenEval][TZ] Add LLM leaderboard for Roblox Studio Assistant in OpenEval open-sour… #2

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[OpenEval][TZ] Add LLM leaderboard for Roblox Studio Assistant in OpenEval open-sour… #2

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
109 changes: 109 additions & 0 deletions LLM_LEADERBOARD.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,109 @@
<!DOCTYPE html>
<html>

<body>

## LLM Leaderboard on Roblox Studio Assistant

<table>
<thead>
<tr>
<th rowspan="2">Model</th>
<th colspan="4" class="eval-pass">Pass Rate </th>
<th colspan="1" class="safety">Safety</th>
<th colspan="2" class="response-behavior">Tool Calling</th>
</tr>
<tr>
<th class="eval-pass"><strong>Pass@1</strong></th>
<th class="eval-pass"><strong>Pass@5</strong></th>
<th class="eval-pass"><strong>Cons@5</strong></th>
<th class="eval-pass"><strong>All@5</strong></th>
<th class="safety"><strong>Safety Pass Rate</strong></th>
<th class="response-behavior"><strong>Avg Tool Error Rate </strong></th>
<th class="response-behavior"><strong>Explanation Rate with Tools</strong></th>
</tr>
</thead>
<tbody>
<tr>
<td class="model-name">Claude-4-sonnet-20250514</td>
<td>55.59%</td>
<td>75.99%</td>
<td>56.68%</td>
<td>34.80%</td>
<td>68.7%</td>
<td>40%</td>
<td>97%</td>
</tr>
<tr>
<td class="model-name">Claude-sonnet-4-5-20250929</td>
<td class="best-score">57.87%</td>
<td>74.43%</td>
<td class="best-score">59.91%</td>
<td class="best-score">38.73%</td>
<td class="best-score">78.0%</td>
<td class="best-score">22%</td>
<td class="best-score">90%</td>
</tr>
<tr>
<td class="model-name">Qwen3-Coder 480B/A35B Instruct</td>
<td>49.61%</td>
<td class="best-score">76.26%</td>
<td>50.56%</td>
<td>23.84%</td>
<td>69.7%</td>
<td>67%</td>
<td>47.30%</td>
</tr>
<tr>
<td class="model-name">GLM 4.6</td>
<td>47.72%</td>
<td>71.15%</td>
<td>48.20%</td>
<td>26.01%</td>
<td>-</td>
<td class="best-score">15%</td>
<td>-</td>
</tr>
<tr>
<td class="model-name">GLM 4.5</td>
<td>41.34%</td>
<td>70.57%</td>
<td>40.18%</td>
<td>18.11%</td>
<td>61.6%</td>
<td>88%</td>
<td>37.6%</td>
</tr>
<tr>
<td class="model-name">Gemini-2.5-pro<br>AUTO thinking, NO web</td>
<td>48.58%</td>
<td>67.00%</td>
<td>49.16%</td>
<td>30.59%</td>
<td>59.60%</td>
<td>40%</td>
<td>92%</td>
</tr>
<tr>
<td class="model-name">Gemini-2.5-flash-preview-09-2025<br>AUTO thinking, NO web</td>
<td>37.24%</td>
<td>63.04%</td>
<td>35.61%</td>
<td>17.72%</td>
<td>68.70%</td>
<td>69%</td>
<td>49.50%</td>
</tr>
</tbody>
</table>
</body>
</html>

## Metrics Explaination
- Pass@1 -- average probability of success in 1 attempt
- Pass@5 -- average probability of success in at least 1 out of 5 attempts
- Cons@5 -- average probability of success in at least 3 out of 5 attempts
- All@5 -- average probability of success in 5 out of 5 attempts
- Safety Pass Rate -- average rate whether it handles safety/appropriateness correctly
- Avg Tool Error Rate -- average tool call error rates
- Explanation Rate with Tools -- quality of explanations when using tools