Telco Troubleshooting Agentic Challenge
€40 000 EUR
18 days left
Agentic AI
Fine-tuning
Large Language Models
886 joined
176 active
Start
Apr 17, 26
Close
May 18, 26
Reveal
May 29, 26
'%3e%3ccircle r='120' fill='%230f47af'/%3e%3cg id='a'%3e%3cpath fill='%23fcdd09' d='m0-96-4.206 12.944 17.348 53.39h-23.13l-2.599 8h74.163l11.011-8H21.553z'/%3e%3cpath stroke='%23fcdd09' stroke-width='4' d='m25.863-35.597 30.564-42.069'/%3e%3c/g%3e%3cuse xlink:href='%23a' transform='rotate(72)'/%3e%3cuse xlink:href='%23a' transform='rotate(144)'/%3e%3cuse xlink:href='%23a' transform='rotate(216)'/%3e%3cuse xlink:href='%23a' transform='rotate(288)'/%3e%3c/g%3e%3c/svg%3e){kind=small}
To the 0.90+ Scorers: Is the data actually this easy, or did anyone fine-tune Qwen?
19 Apr 2026, 06:58 · 3
Huge respect to @yassin104 @Abdourahamane_ @TENGSOABA @d4rk51d3 @Ammart90 @vaderyang @NightJ @cyberkid_001 @snazon @JunbiaoLu @LiQizhe the folks sitting at the top of the leaderboard with 0.90+ and 1.0 scores!
Seeing scores this high on Day 1 makes me wonder if I'm completely overcomplicating things.
I have two questions for the community:
- Is the data really this easy? For the top scorers, are you getting these numbers purely through clean LLM model, or is there trick in the data making it simple?
- Has anyone tried fine-tuning an LLM? . Has anyone successfully used a fine-tuned LLM for this ?
Would love to hear your thoughts! Good luck everyone.
Hey @Yisakberhanu ,
I’m just starting on this challenge and honestly had the exact same confusion from your post. Seeing 0.90+ scores with my below average performance is a bit demoralizing.
I noticed you later broke into the 0.9+ range yourself, which is impressive. Was it mostly better agent design/prompting, or did you realize the data was simpler than it first looked?
Welcome, great man. This time we are competing with perfect gods; we have to compete with gods anyway. My score is heavily invested data, but still I have not even reached perfect. My concern is that perfect in ML is always impossible, even reaching .95, so unless the data contains a leak or the top prefect team has extraordinary power to see what we cannot see, there is no way to score heavily invested data. This is what I can tell you. But we have to learn from the top prefect team. This time I am eager to see the perfect team's solution at the end of the competition. You may reach the perfect score soon. Good luck. Now I feel like I am stupid, okay?
I see. Thanks for the feedback. Let see