You are browsing as a guest. Sign up (or log in) to start making projects!

nish24

@nish24

Joined June 5th, 2026

  • 6Devlogs
  • 2Projects
  • 2Ships
  • 16Votes
Ship Changes requested

OpenMythos is a cybersecurity-focused LLM
It's trained specifically on:
- ArXiv cs.CR (Cryptography & Security) research papers
- Real CVE vulnerability data with detailed descriptions

What did we build?

- Full SFT + RLVR training pipeline on H100 GPUs (via Modal)
- Custom datasets: filtered ArXiv papers + CVE vulnerability data
- Deployed on Hugging Face Spaces

What was challenging?

- Scraping and filtering high-quality dataset
- Getting RLVR to actually improve security reasoning.
- Running stable distributed training on Modal with DeepSpeed
- Making the model actually useful for real vuln analysis, not just theory

What are we proud of?

- Model genuinely understands CVEs and can reason about attack surfaces
- Clean SFT → RLVR pipeline that others can replicate
- 9,000+ users in the first week with zero marketing budget
- Benchmark results on CyberGym and cybench

How to test it?

Hugging Face Space: https://huggingface.co/spaces/build-small-hackathon/OpenMythos

Try prompts like:
- "Analyze this code for security vulnerabilities: [paste your code]"
- "Review this API endpoint for security issues: [paste code]"

It works best for:
✅ Code security review
✅ CVE explanation & mitigation
✅ Attack surface analysis
✅ Security research questions

  • 4 devlogs
  • 50h
Try project → See source code →
Open comments for this post

17m 2s logged

Hit 9,000+ users on OpenMythos last week

OpenMythos is a cybersecurity-focused LLM trained on ArXiv cs.CR papers and real CVE data.

If you’ve tried it:
→ Share your feedback
→ Found a vuln in your project using it? We’d love to hear

Want to stress-test your own project’s security?
Give it a shot 👇
https://huggingface.co/spaces/build-small-hackathon/OpenMythos

0
0
3
Open comments for this post

43h 41m 18s logged

Hey everyone! OpenMythos benchmarks are finally here.

Sorry it took about a week to post these.
The delay was mainly because SWE-bench results weren’t matching up with Qwen 3.6 27B official numbers.

Turns out Qwen used a different eval harness and also refined/filtered the benchmark problems, even there prev 3.5 (72.4 in SWE Verified ) version benchmark score is not matching with the numbers published in 3.6 (75 in SWE Verified).

Anyway, here are the results across SWE-bench Pro, CyberGym, and cybench.

OpenMythos holds up pretty well for a small cybersecurity-focused model!
But it has capability to do better. So, will train it further.

Demo: https://huggingface.co/spaces/build-small-hackathon/OpenMythos
Model: https://huggingface.co/build-small-hackathon/OpenMythos

0
0
8
Ship

I made an Among Use Style Multiplayer Slack Bot game. Where the host can start game with /fn-imposter than players who wanna play clicks join game button to join the game. Only host can start the game if there are less than 4 player the remaining space will be filled by bots. As host starts the game you get your role and tasks in DM than you have to submit the report. Than discussion based on all players submissions that who can be imposter than voting.
If you are imposter you have to try to convince other player you are crewmate and accuse other crewmate and vote him out.
If you are crewmate you have to find the imposter.
The major difficulty I got is in reducing the spamminess and learning the slack api. It's a facinating journey.

Try project → See source code →
Open comments for this post

3h 31m 2s logged

Devlog 2 of OpenMythos

Dataset Progress:

45k rows of dataset completed.
Generation got stuck after 37k due to a bug from upstream provider. Resolved the issue now back on track.

0
0
9

Holy, Can’t believe all this is in ascii

1h 18m 20s logged

First Output test on my ASCIIConverter! Video is compressed so hopefully it plays and looks decent 😭🙏 Detail:220, Mode: Blocky.

1
1
Open comments for this post

47m 42s logged

Devlog 2 of my Among US game

Thanks to Nayte and keerthi for Feedback.

Yeah, So, people have complained that its too much spammy. So, i added threads and editing previous messages to make it less spammy and do not fill any channel with just bot spam.

Feedback and suggestions welcome.

Changelog

0
0
42
Open comments for this post

3h 13m 54s logged

Hey, Everyone
So, I am working on an Among Us style game played entirely inside Slack. Players join a lobby in a channel, get assigned roles (crewmate/impostor), receive mission scenarios via DM, submit observations, discuss and vote to eject the impostor. Bot players (AI) fill empty slots when fewer than 4 humans join.

You can try it in slack using command /fn-imposter
Please try and Give review.

So, currently I have created a basic bot. In next version I want to make game more interesting by adding more roles, more better tasks, way to identify the imposter and more variant of games.

0
0
11

Followers

Loading…