TAPP Security
Của TAPP SecurityWe have built lightweight detectors for Language models, reducing hallucinations by 85% in Gemma-2-9B and doubling the down-stream performance improvement in specialized models, such as CISCO's cybersecurity LLM.
- Giai đoạn
- Bản mẫu
- Ngành
- Ngân hàng và tài chính
- Y tế
- Triển khai
- Đám mây (SaaS)
Giá trị mang lại
Our value proposition is based in a deeply technical moat, originating from graduate level research in pure mathematics. We are unique in our approach and execution. Our internal-based detection methods are 570,000X improvement on current SOTA methods both in terms of compute and data efficiency. We are fundamentally opening a new paradigm for internal-based evaluation in real-time.
Current language models are unreliable and hallucinate often, obstructing their use for high-impact tasks in cybersecurity, finance or healthcare.
Đang mở deck…
Trang 1
Khách hàng phù hợp nhất
Our ideal customers are companies that are developing and deploying their own proprietary small to medium-sized Language models for specialized workflows, such as Harvey, ThomsonReuters, CISCO, Galileo, etc.
Khách hàng tiêu biểu đang trả phí
- We are pre-revenue.
We have a completed an MVP, demonstrated the efficacy of our systems to CISCO and are currently exploring design partnerships.